ts 103 190-1 - v1.3.1 - digital audio compression (ac-4) standard; part 1: channel ... ·...
TRANSCRIPT
ETSI TS 103 190-1 V1.3.1 (2018-02)
Digital Audio Compression (AC-4) Standard; Part 1: Channel based coding
�
TECHNICAL SPECIFICATION
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)2
Reference RTS/JTC-043-1
Keywords audio, broadcasting, codec, content, digital,
distribution
ETSI
650 Route des Lucioles F-06921 Sophia Antipolis Cedex - FRANCE
Tel.: +33 4 92 94 42 00 Fax: +33 4 93 65 47 16
Siret N° 348 623 562 00017 - NAF 742 C
Association à but non lucratif enregistrée à la Sous-Préfecture de Grasse (06) N° 7803/88
Important notice
The present document can be downloaded from: http://www.etsi.org/standards-search
The present document may be made available in electronic versions and/or in print. The content of any electronic and/or print versions of the present document shall not be modified without the prior written authorization of ETSI. In case of any existing or perceived difference in contents between such versions and/or in print, the only prevailing document is
the print of the Portable Document Format (PDF) version kept on a specific network drive within ETSI Secretariat.
Users of the present document should be aware that the document may be subject to revision or change of status. Information on the current status of this and other ETSI documents is available at
http://portal.etsi.org/tb/status/status.asp
If you find errors in the present document, please send your comment to one of the following services: http://portal.etsi.org/chaircor/ETSI_support.asp
Copyright Notification
No part may be reproduced or utilized in any form or by any means, electronic or mechanical, including photocopying and microfilm except as authorized by written permission of ETSI.
The content of the PDF version shall not be modified without the written authorization of ETSI. The copyright and the foregoing restriction extend to reproduction in all media.
© ETSI 2018.
All rights reserved.
DECTTM, PLUGTESTSTM, UMTSTM and the ETSI logo are trademarks of ETSI registered for the benefit of its Members. 3GPPTM and LTE™ are trademarks of ETSI registered for the benefit of its Members and
of the 3GPP Organizational Partners. oneM2M logo is protected for the benefit of its Members.
GSM® and the GSM logo are trademarks registered and owned by the GSM Association.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)3
Contents
Intellectual Property Rights .............................................................................................................................. 17
Foreword ........................................................................................................................................................... 17
Modal verbs terminology .................................................................................................................................. 17
Introduction ...................................................................................................................................................... 18
1 Scope ...................................................................................................................................................... 20
2 References .............................................................................................................................................. 20
2.1 Normative references ....................................................................................................................................... 20
2.2 Informative references ...................................................................................................................................... 20
3 Definitions, symbols, abbreviations and conventions ............................................................................ 21
3.1 Definitions ........................................................................................................................................................ 21
3.2 Symbols ............................................................................................................................................................ 23
3.3 Abbreviations ................................................................................................................................................... 24
3.4 Conventions ...................................................................................................................................................... 25
4 Bitstream syntax ..................................................................................................................................... 26
4.1 Semantics of syntax specification..................................................................................................................... 26
4.2 Syntax specification ......................................................................................................................................... 26
4.2.1 raw_ac4_frame - Raw AC-4 frame ............................................................................................................. 26
4.2.2 variable_bits - Variable bits ........................................................................................................................ 26
4.2.3 AC-4 frame information ............................................................................................................................. 27
4.2.3.1 ac4_toc - AC-4 table of contents ........................................................................................................... 27
4.2.3.2 ac4_presentation_info - AC-4 presentation information ....................................................................... 27
4.2.3.3 presentation_version - Presentation version information ...................................................................... 29
4.2.3.4 frame_rate_multiply_info - Frame rate multiplier information ............................................................. 29
4.2.3.5 emdf_info - EMDF information ............................................................................................................ 29
4.2.3.6 ac4_substream_info - AC-4 substream information .............................................................................. 30
4.2.3.7 content_type - Content type .................................................................................................................. 30
4.2.3.8 presentation_config_ext_info - Presentation configuration extended information ............................... 30
4.2.3.9 ac4_hsf_ext_substream_info - AC-4 HSF extension substream information ....................................... 31
4.2.3.10 emdf_payloads_substream_info - EMDF payloads substream information .......................................... 31
4.2.3.11 substream_index_table - Substream index table ................................................................................... 31
4.2.4 AC-4 substreams ......................................................................................................................................... 32
4.2.4.1 AC-4 substreams ................................................................................................................................... 32
4.2.4.2 ac4_substream - AC-4 substream .......................................................................................................... 32
4.2.4.3 ac4_hsf_ext_substream - AC-4 high sampling frequency extension substream ................................... 32
4.2.4.4 emdf_payloads_substream - EMDF payloads substream ...................................................................... 33
4.2.5 audio_data - Audio data .............................................................................................................................. 33
4.2.6 Channel elements ........................................................................................................................................ 33
4.2.6.1 single_channel_element - Single channel element ................................................................................ 33
4.2.6.2 mono_data - Mono data ........................................................................................................................ 34
4.2.6.3 channel_pair_element - Channel pair element ...................................................................................... 34
4.2.6.4 stereo_data - Stereo data ....................................................................................................................... 35
4.2.6.5 3_0_channel_element - 3.0 channel element......................................................................................... 35
4.2.6.6 5_X_channel_element - 5.X channel element ....................................................................................... 36
4.2.6.7 two_channel_data - Two channel data .................................................................................................. 37
4.2.6.8 three_channel_data - Three channel data .............................................................................................. 37
4.2.6.9 four_channel_data - Four channel data ................................................................................................. 38
4.2.6.10 five_channel_data - Five channel data .................................................................................................. 38
4.2.6.11 three_channel_info - Three channel info............................................................................................... 38
4.2.6.12 four_channel_info - Four channel info .................................................................................................. 38
4.2.6.13 five_channel_info - Five channel info................................................................................................... 39
4.2.6.14 7_X_channel_element - 7.X channel element ....................................................................................... 39
4.2.7 Spectral frontend ......................................................................................................................................... 40
4.2.7.1 sf_info - Spectral frontend info ............................................................................................................. 40
4.2.7.2 sf_info_lfe - Spectral frontend info for LFE ......................................................................................... 40
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)4
4.2.7.3 sf_data - Spectral frontend data ............................................................................................................. 41
4.2.7.4 sf_hsf_data - Spectral frontend HSF extension data ............................................................................. 41
4.2.8 Audio spectral frontend .............................................................................................................................. 41
4.2.8.1 asf_transform_info - ASF transform info .............................................................................................. 41
4.2.8.2 asf_psy_info - ASF scale factor band info ............................................................................................ 41
4.2.8.3 asf_section_data - ASF section data ...................................................................................................... 42
4.2.8.4 asf_spectral_data - ASF spectral data ................................................................................................... 43
4.2.8.5 asf_scalefac_data - ASF scale factor data ............................................................................................. 44
4.2.8.6 asf_snf_data - ASF spectral noise fill data ............................................................................................ 44
4.2.8.7 asf_hsf_spectral_data - ASF spectral data for HSF extension .............................................................. 44
4.2.8.8 asf_hsf_scalefac_data - ASF scale factor data for HSF extension ........................................................ 45
4.2.8.9 asf_hsf_snf_data - ASF spectral noise fill data for HSF extension ....................................................... 45
4.2.9 Speech spectral frontend ............................................................................................................................. 46
4.2.9.1 ssf_data - Speech spectral frontend data ............................................................................................... 46
4.2.9.2 ssf_granule - Speech spectral frontend granule ..................................................................................... 46
4.2.9.3 ssf_st_data - Speech spectral frontend static data ................................................................................. 47
4.2.9.4 ssf_ac_data - Speech spectral frontend arithmetic coded data .............................................................. 47
4.2.10 Stereo audio processing .............................................................................................................................. 48
4.2.10.1 chparam_info -Stereo information ........................................................................................................ 48
4.2.10.2 sap_data - Stereo audio processing data ................................................................................................ 48
4.2.11 Companding control ................................................................................................................................... 49
4.2.12 Advanced spectral extension - A-SPX ........................................................................................................ 49
4.2.12.1 aspx_config - A-SPX configuration ...................................................................................................... 49
4.2.12.2 aspx_data_1ch - A-SPX 1-channel data ................................................................................................ 49
4.2.12.3 aspx_data_2ch - A-SPX 2-channel data ................................................................................................ 50
4.2.12.4 aspx_framing - A-SPX framing ............................................................................................................ 50
4.2.12.5 aspx_delta_dir - A-SPX direction of envelope delta coding ................................................................. 51
4.2.12.6 aspx_hfgen_iwc_1ch - A-SPX 1-channel HF generation and interleaved waveform coding ............... 52
4.2.12.7 aspx_hfgen_iwc_2ch - A-SPX 2-channel HF generation and interleaved waveform coding ............... 52
4.2.12.8 aspx_ec_data - A-SPX entropy coded data ........................................................................................... 53
4.2.12.9 aspx_huff_data - A-SPX Huffman data ................................................................................................ 54
4.2.13 Advanced coupling - A-CPL ...................................................................................................................... 54
4.2.13.1 acpl_config_1ch - A-CPL 1-channel configuration .............................................................................. 54
4.2.13.2 acpl_config_2ch - A-CPL 2-channel configuration .............................................................................. 54
4.2.13.3 acpl_data_1ch - A-CPL 1-channel data ................................................................................................. 55
4.2.13.4 acpl_data_2ch - A-CPL 2-channel data ................................................................................................. 55
4.2.13.5 acpl_framing_data - A-CPL framing data ............................................................................................. 55
4.2.13.6 acpl_ec_data - A-CPL entropy coded data ............................................................................................ 56
4.2.13.7 acpl_huff_data - A-CPL Huffman data ................................................................................................. 56
4.2.14 Metadata ..................................................................................................................................................... 56
4.2.14.1 metadata() - Metadata ........................................................................................................................... 56
4.2.14.2 basic_metadata - Basic metadata ........................................................................................................... 57
4.2.14.3 further_loudness_info - Additional loudness information ..................................................................... 58
4.2.14.4 extended_metadata - Extended metadata .............................................................................................. 59
4.2.14.5 drc_frame - DRC frame ........................................................................................................................ 60
4.2.14.6 drc_config - DRC configuration ........................................................................................................... 60
4.2.14.7 drc_decoder_mode_config - DRC decoder mode configuration ........................................................... 60
4.2.14.8 drc_compression_curve - Compression curve parameters .................................................................... 61
4.2.14.9 drc_data -DRC frame-based data .......................................................................................................... 61
4.2.14.10 drc_gains - DRC gains .......................................................................................................................... 62
4.2.14.11 dialog_enhancement - Dialog enhancement metadata .......................................................................... 62
4.2.14.12 de_config - Dialog enhancement configuration .................................................................................... 62
4.2.14.13 de_data - Dialog enhancement data ...................................................................................................... 63
4.2.14.14 emdf_payload_config - EMDF payload configuration ......................................................................... 64
4.2.14.15 emdf_protection - EMDF protection data ............................................................................................. 64
4.3 Description of bitstream elements .................................................................................................................... 64
4.3.0 Introduction................................................................................................................................................. 64
4.3.1 raw_ac4_frame - raw AC-4 frame .............................................................................................................. 65
4.3.1.1 fill_area - fill area .................................................................................................................................. 65
4.3.1.2 fill_bits - byte alignment bits ................................................................................................................ 65
4.3.1.3 byte_align - byte alignment bits ............................................................................................................ 65
4.3.2 variable_bits - Variable bits ........................................................................................................................ 65
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)5
4.3.2.0 Encoding ............................................................................................................................................... 65
4.3.2.1 read - read bits ....................................................................................................................................... 65
4.3.2.2 b_read_more - read more flag ............................................................................................................... 65
4.3.3 AC-4 frame information ............................................................................................................................. 66
4.3.3.1 Purpose .................................................................................................................................................. 66
4.3.3.2 ac4_toc - AC-4 table of contents ........................................................................................................... 66
4.3.3.2.1 bitstream_version - bitstream version .............................................................................................. 66
4.3.3.2.2 sequence_counter - sequence counter .............................................................................................. 66
4.3.3.2.3 b_wait_frames ................................................................................................................................. 66
4.3.3.2.4 wait_frames ..................................................................................................................................... 66
4.3.3.2.5 fs_index - Sampling frequency index .............................................................................................. 67
4.3.3.2.6 frame_rate_index - frame rate ......................................................................................................... 67
4.3.3.2.7 b_iframe_global - global I-frame flag ............................................................................................. 68
4.3.3.2.8 b_single_presentation - single presentation flag .............................................................................. 68
4.3.3.2.9 b_more_presentations - more presentations flag ............................................................................. 68
4.3.3.2.10 b_payload_base - payload base flag ................................................................................................ 68
4.3.3.2.11 payload_base_minus1 - payload base offset minus 1 ...................................................................... 69
4.3.3.2.12 byte_align - byte alignment bits ...................................................................................................... 69
4.3.3.3 ac4_presentation_info - AC-4 presentation information ....................................................................... 69
4.3.3.3.1 b_single_substream - single substream flag .................................................................................... 69
4.3.3.3.2 b_belongs_to_presentation_id - Presentation identifier assignment flag......................................... 69
4.3.3.3.3 b_hsf_ext - high sampling frequency extension flag ....................................................................... 69
4.3.3.3.4 presentation_config - presentation configuration - .......................................................................... 69
4.3.3.3.5 b_pre_virtualized - pre-virtualized flag ........................................................................................... 69
4.3.3.3.6 b_add_emdf_substreams - additional EMDF substreams flag ........................................................ 69
4.3.3.3.7 n_add_emdf_substreams - number of additional EMDF substreams .............................................. 69
4.3.3.3.8 mdcompat - compatibility indication ............................................................................................... 70
4.3.3.4 presentation_version - presentation version information ...................................................................... 70
4.3.3.4.1 b_tmp - temporary flag .................................................................................................................... 70
4.3.3.5 frame_rate_multiply_info - frame rate multiplier information .............................................................. 70
4.3.3.5.1 b_multiplier - multiplier flag ........................................................................................................... 70
4.3.3.5.2 multiplier_bit - multiplier bit ........................................................................................................... 70
4.3.3.5.3 frame_rate_factor - frame rate factor ............................................................................................... 70
4.3.3.6 emdf_info - EMDF information ............................................................................................................ 71
4.3.3.6.1 emdf_version - EMDF syntax version ............................................................................................. 71
4.3.3.6.2 key_id - authentication ID ............................................................................................................... 71
4.3.3.6.3 b_emdf_payloads_substream_info - EMDF payloads substream information flag ......................... 71
4.3.3.7 ac4_substream_info - AC-4 substream information .............................................................................. 71
4.3.3.7.1 channel_mode - channel mode ........................................................................................................ 71
4.3.3.7.2 b_sf_multiplier - sampling frequency multiplier flag ...................................................................... 71
4.3.3.7.3 sf_multiplier - sampling frequency multiplier bit ............................................................................ 72
4.3.3.7.4 b_bitrate_info - bit-rate presence flag .............................................................................................. 72
4.3.3.7.5 bitrate_indicator - bit-rate indicator ................................................................................................. 72
4.3.3.7.6 add_ch_base - additional channels coupling base ........................................................................... 72
4.3.3.7.7 b_content_type - content type presence flag.................................................................................... 72
4.3.3.7.8 b_iframe - I-frame flag .................................................................................................................... 72
4.3.3.7.9 substream_index - substream index ................................................................................................. 72
4.3.3.8 content_type - content type ................................................................................................................... 73
4.3.3.8.1 content_classifier - content classifier ............................................................................................... 73
4.3.3.8.2 b_language_indicator - programme language indicator flag ........................................................... 73
4.3.3.8.3 b_serialized_language_tag - serialized language tag flag ................................................................ 73
4.3.3.8.4 b_start_tag - language tag start flag ................................................................................................. 73
4.3.3.8.5 language_tag_chunk - language tag chunk ...................................................................................... 73
4.3.3.8.6 n_language_tag_bytes - number of language tag bytes ................................................................... 74
4.3.3.8.7 language_tag_bytes - language tag bytes......................................................................................... 74
4.3.3.9 presentation_config_ext_info - presentation configuration extended information ................................ 74
4.3.3.9.1 n_skip_bytes - number of bytes to skip ........................................................................................... 74
4.3.3.9.2 b_more_skip_bytes - more bytes to skip flag .................................................................................. 74
4.3.3.9.3 reserved - Reserved ......................................................................................................................... 74
4.3.3.10 ac4_hsf_ext_substream_info - AC-4 HSF extension substream information ....................................... 74
4.3.3.10.1 substream_index - substream index ................................................................................................. 74
4.3.3.11 emdf_payloads_substream_info - EMDF payloads substream information .......................................... 74
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)6
4.3.3.11.1 substream_index - substream index ................................................................................................. 74
4.3.3.12 substream_index_table - substream index table .................................................................................... 74
4.3.3.12.1 n_substreams - number of substreams ............................................................................................. 74
4.3.3.12.2 b_size_present - size present flag .................................................................................................... 74
4.3.3.12.3 b_more_bits - more bits flag ............................................................................................................ 75
4.3.3.12.4 substream_size - substream size ...................................................................................................... 75
4.3.4 ac4_substream - AC-4 substream ............................................................................................................... 75
4.3.4.1 audio_size_value - audio size value ...................................................................................................... 75
4.3.4.2 b_more_bits - more bits flag ................................................................................................................. 75
4.3.4.3 byte_align - byte alignment bits ............................................................................................................ 75
4.3.5 Channel elements ........................................................................................................................................ 75
4.3.5.0 Introduction ........................................................................................................................................... 75
4.3.5.1 mono_codec_mode - mono codec mode ............................................................................................... 75
4.3.5.2 spec_frontend - spectral frontend selection ........................................................................................... 76
4.3.5.3 stereo_codec_mode - stereo codec mode .............................................................................................. 76
4.3.5.4 3_0_codec_mode - 3.0 codec mode ...................................................................................................... 76
4.3.5.5 3_0_coding_config - 3.0 coding configuration ..................................................................................... 76
4.3.5.6 5_X_codec_mode - 5.X codec mode .................................................................................................... 76
4.3.5.7 7_X_codec_mode - 7.X codec mode .................................................................................................... 77
4.3.5.8 coding_config - coding configuration ................................................................................................... 77
4.3.5.9 2ch_mode - channel coupling mode ...................................................................................................... 77
4.3.5.10 b_enable_mdct_stereo_proc - enable MDCT stereo processing flag .................................................... 77
4.3.5.11 chel_matsel - matrix selection code ...................................................................................................... 77
4.3.5.12 b_use_sap_add_ch - use SAP for additional channels flag ................................................................... 77
4.3.5.13 max_sfb_master - max_sfb indication for related channels .................................................................. 77
4.3.6 audio spectral front end ............................................................................................................................... 77
4.3.6.1 asf_transform_info - ASF transform information ................................................................................. 77
4.3.6.1.1 b_long_frame - long frame flag ....................................................................................................... 77
4.3.6.1.2 transf_length[i] - transform length index i ....................................................................................... 78
4.3.6.1.3 transf_length - transform length ...................................................................................................... 78
4.3.6.1.4 get_transf_length(g) – get transform length for group g ................................................................. 79
4.3.6.2 asf_psy_info - ASF psy information ..................................................................................................... 80
4.3.6.2.1 n_msfb_bits - number of maxsfb bits .............................................................................................. 80
4.3.6.2.2 max_sfb[i] - number of transmitted scale factor bands for index i .................................................. 81
4.3.6.2.3 max_sfb_side[i] - number of transmitted scale factor bands for side channel and index i .............. 81
4.3.6.2.4 n_grp_bits - number of grouping bits .............................................................................................. 81
4.3.6.2.5 scale_factor_grouping_bit - scale factor grouping bit ..................................................................... 82
4.3.6.2.6 asf_psy_info helper elements .......................................................................................................... 82
4.3.6.2.7 get_max_sfb(g) – get max_sfb for group g ..................................................................................... 83
4.3.6.3 asf_section_data - ASF section data ...................................................................................................... 83
4.3.6.3.1 sect_cb[g][i] - section codebook ...................................................................................................... 83
4.3.6.3.2 sect_len_incr - section length increment ......................................................................................... 84
4.3.6.4 asf_spectral_data - ASF spectral data ................................................................................................... 84
4.3.6.4.1 asf_qspec_hcw - Huffman coded quantized spectral lines .............................................................. 84
4.3.6.4.2 huff_decode(hcb, hcw) - Huffman decoding ................................................................................... 84
4.3.6.4.3 quad_sign_bits - quad sign bits ....................................................................................................... 84
4.3.6.4.4 pair_sign_bits - pair sign bits ........................................................................................................... 84
4.3.6.4.5 ext_code - extension code................................................................................................................ 84
4.3.6.5 asf_scalefac_data - ASF scale factor data ............................................................................................. 84
4.3.6.5.1 reference_scale_factor - reference scale factor ................................................................................ 84
4.3.6.5.2 asf_sf_hcw - Huffman coded scale factor delta ............................................................................... 84
4.3.6.6 asf_snf_data - ASF spectral noise fill data ............................................................................................ 84
4.3.6.6.1 b_snf_data_exists - spectral noise fill data exists flag ..................................................................... 84
4.3.6.6.2 asf_snf_hcw - Huffman code spectral noise fill delta ...................................................................... 85
4.3.7 Speech spectral frontend ............................................................................................................................. 85
4.3.7.1 ssf_data - Speech spectral frontend data ............................................................................................... 85
4.3.7.1.1 b_ssf_iframe - SSF I-frame flag ...................................................................................................... 85
4.3.7.2 ssf_granule - speech spectral frontend granule ..................................................................................... 85
4.3.7.2.1 stride_flag - stride flag ..................................................................................................................... 85
4.3.7.2.2 SSF configuration ............................................................................................................................ 85
4.3.7.2.3 num_bands_minus12 - number of SSF coded bands minus 12 ....................................................... 86
4.3.7.2.4 predictor_presence_flag[b] - predictor presence flag for block b .................................................... 86
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)7
4.3.7.2.5 delta_flag[b] - delta coding flag for block b .................................................................................... 86
4.3.7.3 ssf_st_data - speech spectral frontend static data .................................................................................. 86
4.3.7.3.1 env_curr_band0_bits - signal envelope index for band 0 ................................................................ 86
4.3.7.3.2 env_startup_band0_bits - startup envelope index for band 0 .......................................................... 86
4.3.7.3.3 gain_bits[b] - envelope gain bits for block b ................................................................................... 86
4.3.7.3.4 predictor_lag_delta_bits[b] - predictor lag delta for block b ........................................................... 86
4.3.7.3.5 predictor_lag_bits[b] - predictor lag index for block b .................................................................... 86
4.3.7.3.6 variance_preserving_flag[b] - variance preserving flag for block b ................................................ 86
4.3.7.3.7 alloc_offset_bits[block] - allocation offset bits for block b ............................................................. 86
4.3.7.4 ssf_ac_data - speech spectral frontend arithmetic coded data ............................................................... 87
4.3.7.4.1 env_curr_ac_bits - arithmetic coded signal envelope indices .......................................................... 87
4.3.7.4.2 env_startup_ac_bits - arithmetic coded startup envelope indices .................................................... 87
4.3.7.4.3 predictor_gain_ac_bits[b] - arithmetic coded predictor gain index for block b ............................... 87
4.3.7.4.4 q_mdct_coefficients_ac_bits[b] - arithmetic coded quantized MDCT coefficients for block b ...... 87
4.3.7.5 SSF helper elements .............................................................................................................................. 87
4.3.8 Stereo audio processing .............................................................................................................................. 87
4.3.8.1 chparam_info - stereo information ........................................................................................................ 87
4.3.8.1.1 sap_mode - stereo audio processing mode ...................................................................................... 87
4.3.8.1.2 ms_used - M/S coding used ............................................................................................................. 88
4.3.8.2 sap_data - stereo audio processing data ................................................................................................ 88
4.3.8.2.1 sap_coeff_all - SAP coding all scale factor bands flag .................................................................... 88
4.3.8.2.2 sap_coeff_used - SAP coding used in scale factor band sfb ............................................................ 88
4.3.8.2.3 delta_code_time - delta coding in time ............................................................................................ 88
4.3.8.2.4 sap_hcw - Huffman coded alpha_q delta ......................................................................................... 88
4.3.9 Companding control ................................................................................................................................... 88
4.3.9.1 sync_flag ............................................................................................................................................... 88
4.3.9.2 b_compand_on ...................................................................................................................................... 88
4.3.9.3 b_compand_avg .................................................................................................................................... 89
4.3.10 Advanced spectral extension - A-SPX ........................................................................................................ 89
4.3.10.1 aspx_config - A-SPX configuration ...................................................................................................... 89
4.3.10.1.1 aspx_quant_mode_env .................................................................................................................... 89
4.3.10.1.2 aspx_start_freq - A-SPX start QMF subband .................................................................................. 89
4.3.10.1.3 aspx_stop_freq - A-SPX stop QMF subband .................................................................................. 89
4.3.10.1.4 aspx_master_freq_scale - A-SPX master frequency table scale ...................................................... 89
4.3.10.1.5 aspx_interpolation - A-SPX interpolation used ............................................................................... 90
4.3.10.1.6 aspx_preflat - A-SPX pre-flattening used ........................................................................................ 90
4.3.10.1.7 aspx_limiter - A-SPX limiter used .................................................................................................. 90
4.3.10.1.8 aspx_noise_sbg - A-SPX number of noise subband groups ............................................................ 90
4.3.10.1.9 aspx_num_env_bits_fixfix - A-SPX frame class FIXFIX bit count ................................................ 90
4.3.10.1.10 aspx_freq_res_mode - A-SPX frequency resolution transmission mode ........................................ 91
4.3.10.2 aspx_data_1ch - A-SPX 1-channel data ................................................................................................ 91
4.3.10.2.1 aspx_xover_subband_offset - A-SPX crossover subband offset ..................................................... 91
4.3.10.3 aspx_data_2ch - A-SPX 2-channel data ................................................................................................ 91
4.3.10.3.1 aspx_xover_subband_offset - A-SPX crossover subband offset ..................................................... 91
4.3.10.3.2 aspx_balance - A-SPX balance setting ............................................................................................ 91
4.3.10.4 aspx_framing - A-SPX framing ............................................................................................................ 91
4.3.10.4.1 aspx_int_class - A-SPX interval class ............................................................................................. 91
4.3.10.4.2 tmp_num_env - temporary variable................................................................................................. 92
4.3.10.4.3 aspx_freq_res[ch][env] - frequency resolution ................................................................................ 92
4.3.10.4.4 aspx_var_bord_left[ch] - leading VAR interval envelope border ................................................... 93
4.3.10.4.5 aspx_var_bord_right[ch] - trailing VAR interval envelope border - 2-bits ..................................... 93
4.3.10.4.6 aspx_num_rel_left[ch] - relative envelope border - 1 or 2 bits ........................................................ 93
4.3.10.4.7 aspx_num_rel_right[ch] - relative envelope border - 1 or 2 bits ..................................................... 93
4.3.10.4.8 aspx_rel_bord_left[ch][rel] - leading relative envelope borders ...................................................... 93
4.3.10.4.9 aspx_rel_bord_right[ch][rel] - trailing relative envelope borders .................................................... 93
4.3.10.4.10 aspx_tsg_ptr[ch] - pointer to envelope border ................................................................................. 93
4.3.10.4.11 A-SPX framing helper variables ...................................................................................................... 93
4.3.10.5 aspx_delta_dir - A-SPX delta coding direction ..................................................................................... 94
4.3.10.5.1 aspx_sig_delta_dir[ch][env] - A-SPX delta coding for signal envelopes ........................................ 94
4.3.10.5.2 aspx_noise_delta_dir[ch][env] - A-SPX delta coding for noise envelopes ..................................... 94
4.3.10.6 aspx_hfgen_iwc_1ch - A-SPX 1-channel HF generation and interleaved waveform coding ............... 94
4.3.10.6.1 aspx_tna_mode[n] - A-SPX subband tonal to noise ratio adjustment mode ................................... 94
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)8
4.3.10.6.2 aspx_add_harmonic[n] - A-SPX add harmonics ............................................................................. 95
4.3.10.6.3 aspx_fic_present - A-SPX frequency interleaved coding present ................................................... 95
4.3.10.6.4 aspx_fic_used_in_sfb[n] - A-SPX frequency interleaved coding used in subband group ............... 95
4.3.10.6.5 aspx_tic_present - A-SPX time interleaved coding present............................................................. 95
4.3.10.6.6 aspx_tic_used_in_slot[n] - A-SPX time interleaved coding used in slot ......................................... 95
4.3.10.6.7 aspx_ah_present - A-SPX add harmonics present ........................................................................... 95
4.3.10.7 aspx_hfgen_iwc_2ch - A-SPX 2-channel HF generation and interleaved waveform coding ............... 96
4.3.10.7.1 aspx_tna_mode[ch][n] - A-SPX subband tonal to noise adjustment mode ..................................... 96
4.3.10.7.2 aspx_add_harmonic[ch][n] - A-SPX add harmonics ....................................................................... 96
4.3.10.7.3 aspx_fic_present - A-SPX frequency interleaved coding present - 1 bit ......................................... 96
4.3.10.7.4 aspx_fic_left - A-SPX frequency interleaved coding in left channel .............................................. 96
4.3.10.7.5 aspx_fic_right - A-SPX frequency interleaved coding in right channel .......................................... 96
4.3.10.7.6 aspx_fic_used_in_sfb[ch][n] - A-SPX frequency interleaved coding used in subband group ........ 96
4.3.10.7.7 aspx_tic_present - A-SPX time interleaved coding present............................................................. 96
4.3.10.7.8 aspx_tic_copy - A-SPX time interleaved coding copy data ............................................................ 96
4.3.10.7.9 aspx_tic_left - A-SPX time interleaved coding in left channel ....................................................... 97
4.3.10.7.10 aspx_tic_right - A-SPX time interleaved coding in right channel ................................................... 97
4.3.10.7.11 aspx_tic_used_in_slot[n] - A-SPX time interleaved coding used in slots ....................................... 97
4.3.10.7.12 aspx_ah_left - A-SPX add harmonics in left channel ...................................................................... 97
4.3.10.7.13 aspx_ah_right - A-SPX add harmonics in right channel ................................................................. 97
4.3.10.8 Functions for Huffman coding .............................................................................................................. 97
4.3.10.8.0 aspx_ec_data - A-SPX Huffman data .............................................................................................. 97
4.3.10.8.1 aspx_huff_data - Huffman decoding for A-SPX values .................................................................. 98
4.3.10.8.2 aspx_hcw - A-SPX Huffman code word ......................................................................................... 98
4.3.10.8.3 huff_decode_diff(hcb, hcw) - Huffman decoding for differences ................................................... 98
4.3.11 Advanced coupling - A-CPL ...................................................................................................................... 98
4.3.11.1 acpl_config_1ch - A-CPL 1-channel configuration .............................................................................. 98
4.3.11.1.1 acpl_1ch_mode - A-CPL tool 1-channel mode ............................................................................... 98
4.3.11.1.2 acpl_num_param_bands_id - A-CPL number of parameter bands .................................................. 98
4.3.11.1.3 acpl_quant_mode - A-CPL quantization mode ............................................................................... 98
4.3.11.1.4 acpl_qmf_band_minus1 - A-CPL QMF band minus 1 .................................................................... 99
4.3.11.2 acpl_config_2ch - A-CPL 2-channel configuration .............................................................................. 99
4.3.11.2.1 acpl_num_param_bands_id - A-CPL number of parameter bands .................................................. 99
4.3.11.2.2 acpl_quant_mode_0 - A-CPL quantization mode 0......................................................................... 99
4.3.11.2.3 acpl_quant _mode_1 - A-CPL quantization mode 1........................................................................ 99
4.3.11.3 acpl_data_1ch - A-CPL 1-channel data ................................................................................................. 99
4.3.11.4 acpl_data_2ch - A-CPL 2-channel data ................................................................................................. 99
4.3.11.5 acpl_framing_data - A-CPL framing data ............................................................................................. 99
4.3.11.5.1 acpl_interpolation_type - A-CPL interpolation type ....................................................................... 99
4.3.11.5.2 acpl_num_param_sets_cod - A-CPL number of parameter sets per frame ..................................... 99
4.3.11.5.3 acpl_param_timeslot - A-CPL parameter change at QMF time slot.............................................. 100
4.3.11.6 acpl_huff_data - A-CPL Huffman data ............................................................................................... 100
4.3.11.6.1 acpl_hcw - A-CPL Huffman code word ........................................................................................ 100
4.3.12 Basic and extended metadata .................................................................................................................... 100
4.3.12.1 metadata - metadata............................................................................................................................. 100
4.3.12.1.1 tools_metadata_size_value - size of tools metadata ...................................................................... 100
4.3.12.1.2 b_more_bits - more bits flag .......................................................................................................... 100
4.3.12.1.3 b_emdf_payloads_substream - EMDF payloads substream flag ................................................... 100
4.3.12.2 basic_metadata - basic metadata ......................................................................................................... 100
4.3.12.2.1 dialnorm_bits - input reference level ............................................................................................. 100
4.3.12.2.2 b_more_basic_metadata - more basic metadata flag ..................................................................... 101
4.3.12.2.3 b_further_loudness_info - additional loudness information flag ................................................... 101
4.3.12.2.4 b_prev_dmx_info - previous downmix information flag ............................................................... 101
4.3.12.2.5 pre_dmixtyp_2ch - previous downmix to 2 channels type ............................................................ 101
4.3.12.2.6 phase90_info_2ch - phase 90 in 2 channels info ........................................................................... 101
4.3.12.2.7 b_dmx_coeff - downmix coefficients present flag ........................................................................ 101
4.3.12.2.8 loro_center_mixgain - LoRo Centre mix gain ............................................................................... 101
4.3.12.2.9 loro_surround_mixgain - LoRo surround mix gain ....................................................................... 102
4.3.12.2.10 b_loro_dmx_loud_corr - LoRo downmix loudness correction flag ............................................... 102
4.3.12.2.11 loro_dmx_loud_corr - LoRo downmix loudness correction .......................................................... 102
4.3.12.2.12 b_ltrt_mixinfo - LtRt downmix information present ..................................................................... 102
4.3.12.2.13 ltrt_center_mixgain - LtRt Centre mix gain .................................................................................. 102
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)9
4.3.12.2.14 ltrt_surround_mixgain - LtRt surround mix gain .......................................................................... 102
4.3.12.2.15 b_ltrt_dmx_loud_corr - LtRt downmix loudness correction flag .................................................. 102
4.3.12.2.16 ltrt_dmx_loud_corr - LtRt downmix loudness correction ............................................................. 102
4.3.12.2.17 b_lfe_mixinfo - LFE downmix information present ...................................................................... 103
4.3.12.2.18 lfe_mixgain - LFE mix gain .......................................................................................................... 103
4.3.12.2.19 preferred_dmx_method - preferred downmix method ................................................................... 103
4.3.12.2.20 b_predmxtyp_5ch - previous downmix to five channels flag ........................................................ 103
4.3.12.2.21 pre_dmixtyp_5ch - previous downmix to five channels type ........................................................ 103
4.3.12.2.22 b_preupmixtyp_5ch - previous upmix to five channels flag.......................................................... 104
4.3.12.2.23 pre_upmixtyp_5ch - previous upmix to five channels type ........................................................... 104
4.3.12.2.24 b_upmixtyp_7ch - previous upmix to seven channels flag ............................................................ 104
4.3.12.2.25 pre_upmixtyp_3_4 - previous upmix to seven channels type ........................................................ 104
4.3.12.2.26 pre_upmixtyp_3_2_2 - previous upmix to seven channels type .................................................... 104
4.3.12.2.27 phase90_info_mc - phase 90 in multi-channel information .......................................................... 104
4.3.12.2.28 b_surround_attenuation_known - surround attenuation known flag ............................................. 105
4.3.12.2.29 b_lfe_attenuation_known - LFE attenuation known flag .............................................................. 105
4.3.12.2.30 b_dc_blocking - DC blocking flag ................................................................................................ 105
4.3.12.2.31 dc_block_on - DC blocking ........................................................................................................... 105
4.3.12.3 further_loudness_info - additional loudness information .................................................................... 105
4.3.12.3.1 loudness_version - loudness version ............................................................................................. 105
4.3.12.3.2 extended_loudness_version - loudness version extension ............................................................. 105
4.3.12.3.3 loud_prac_type - loudness practice type ........................................................................................ 105
4.3.12.3.4 b_loudcorr_dialgate - loudness correction dialogue gating flag .................................................... 105
4.3.12.3.5 dialgate_prac_type - dialogue gating practice type ....................................................................... 106
4.3.12.3.6 b_loudcorr_type - loudness correction type ................................................................................... 106
4.3.12.3.7 b_loudrelgat - loudness value relative gated flag .......................................................................... 106
4.3.12.3.8 loudrelgat - loudness value relative gated...................................................................................... 106
4.3.12.3.9 b_loudspchgat - loudness value speech gated flag ........................................................................ 106
4.3.12.3.10 loudspchgat - loudness value speech gated.................................................................................... 106
4.3.12.3.11 dialgate_prac_type - dialogue gating practice type ....................................................................... 107
4.3.12.3.12 b_loudstrm3s - loudness values short term 3s flag ........................................................................ 107
4.3.12.3.13 loudstrm3s - loudness values short term 3s ................................................................................... 107
4.3.12.3.14 b_max_loudstrm3s - maximum loudness value short term 3s flag ................................................ 107
4.3.12.3.15 max_loudstrm3s - maximum loudness value short term 3s ........................................................... 107
4.3.12.3.16 b_truepk - true peak flag ................................................................................................................ 107
4.3.12.3.17 truepk - true peak ........................................................................................................................... 107
4.3.12.3.18 b_max_truepk - maximum true peak flag ...................................................................................... 108
4.3.12.3.19 max_truepk - maximum true peak ................................................................................................. 108
4.3.12.3.20 b_prgmbndy - programme boundary flag ...................................................................................... 108
4.3.12.3.21 prgmbndy_bit - programme boundary bit ...................................................................................... 108
4.3.12.3.22 b_end_or_start - programme boundary end or start flag ............................................................... 108
4.3.12.3.23 b_prgmbndy_offset - programme boundary offset flag ................................................................. 108
4.3.12.3.24 prgmbndy_offset - programme boundary offset ............................................................................ 108
4.3.12.3.25 b_lra - loudness range flag............................................................................................................. 108
4.3.12.3.26 lra - loudness range ........................................................................................................................ 108
4.3.12.3.27 lra_prac_type - loudness range measurement practice type........................................................... 109
4.3.12.3.28 b_loudmntry - momentary loudness flag ....................................................................................... 109
4.3.12.3.29 loudmntry - momentary loudness .................................................................................................. 109
4.3.12.3.30 b_max_loudmntry - maximum momentary loudness flag ............................................................. 109
4.3.12.3.31 max_loudmntry - maximum momentary loudness ........................................................................ 109
4.3.12.3.32 b_extension - extension flag .......................................................................................................... 109
4.3.12.3.33 e_bits_size - extension size............................................................................................................ 109
4.3.12.3.34 extension_bits - extension bits ....................................................................................................... 109
4.3.12.4 extended_metadata .............................................................................................................................. 110
4.3.12.4.1 b_associated - associate substream flag - parameter...................................................................... 110
4.3.12.4.2 b_dialog - dialogue substream flag - parameter ............................................................................. 110
4.3.12.4.3 b_scale_main - scale main flag ...................................................................................................... 110
4.3.12.4.4 scale_main - scale main ................................................................................................................. 110
4.3.12.4.5 b_scale_main_center - scale main centre flag ............................................................................... 110
4.3.12.4.6 scale_main_center - scale main centre .......................................................................................... 110
4.3.12.4.7 b_scale_main_front - scale main front flag ................................................................................... 110
4.3.12.4.8 scale_main_front - scale main front .............................................................................................. 111
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)10
4.3.12.4.9 pan_associated - associate pan data- 8 bits .................................................................................... 111
4.3.12.4.10 b_dialog_max_gain - dialog maximum gain flag .......................................................................... 111
4.3.12.4.11 dialog_max_gain - dialogue maximum gain ................................................................................. 111
4.3.12.4.12 b_pan_dialog_present - dialogue pan data present flag ................................................................. 111
4.3.12.4.13 pan_dialog - dialogue pan data ...................................................................................................... 111
4.3.12.4.14 pan_signal_selector - selector of panning signal .......................................................................... 111
4.3.12.4.15 b_channels_classifier - channel classifier flag ............................................................................... 112
4.3.12.4.16 b_{c,l,r,ls,rs,lrs,rrs,lw,rw,vhl,vhr,lfe}_active - channel active flag ............................................... 112
4.3.12.4.17 b_{c,l,r}_has_dialog - channel has dialogue flag .......................................................................... 112
4.3.12.4.18 b_event_probability_present - event probability present flag........................................................ 112
4.3.12.4.19 event_probability - event probability ............................................................................................. 112
4.3.12.5 Channel mode query functions ............................................................................................................ 112
4.3.12.5.1 channel_mode_contains_Lfe() ...................................................................................................... 112
4.3.12.5.2 channel_mode_contains_c() .......................................................................................................... 112
4.3.12.5.3 channel_mode_contains_lr() ......................................................................................................... 112
4.3.12.5.4 channel_mode_contains_LsRs() .................................................................................................... 113
4.3.12.5.5 channel_mode_contains_LrsRrs() ................................................................................................. 113
4.3.12.5.6 channel_mode_contains_LwRw() ................................................................................................. 113
4.3.12.5.7 channel_mode_contains_VhlVhr() ................................................................................................ 113
4.3.13 Dynamic range control - DRC .................................................................................................................. 114
4.3.13.1 drc_frame - dynamic range control ..................................................................................................... 114
4.3.13.1.1 b_drc_present - DRC present ........................................................................................................ 114
4.3.13.2 drc_config - DRC configuration ......................................................................................................... 114
4.3.13.2.1 drc_decoder_nr_modes - number of DRC decoder modes ............................................................ 114
4.3.13.2.2 drc_eac3_profile - (E-)AC-3 profile .............................................................................................. 114
4.3.13.3 drc_decoder_mode_config - DRC decoder mode configuration ......................................................... 114
4.3.13.3.1 drc_decoder_mode_id - DRC decoder mode ID ........................................................................... 114
4.3.13.3.2 drc_output_level_from - lowest reference output level ................................................................. 114
4.3.13.3.3 drc_output_level_to - highest reference output level ..................................................................... 115
4.3.13.3.4 drc_repeat_profile_flag - repeat profile flag .................................................................................. 115
4.3.13.3.5 drc_repeat_id - repeat data from ID............................................................................................... 115
4.3.13.3.6 drc_default_profile_flag - default profile flag ............................................................................... 115
4.3.13.3.7 drc_compression_curve_flag - compression curve flag ................................................................ 117
4.3.13.3.8 drc_gains_config - DRC gains configuration ................................................................................ 117
4.3.13.4 drc_compression_curve - compression curve parameters ................................................................... 118
4.3.13.4.1 DRC compression curve parameterization .................................................................................... 118
4.3.13.4.2 drc_lev_nullband_low - null band lower boundary ....................................................................... 118
4.3.13.4.3 drc_lev_nullband_high - null band higher boundary ..................................................................... 118
4.3.13.4.4 drc_gain_max_boost - maximum boost ......................................................................................... 118
4.3.13.4.5 drc_lev_max_boost - start of maximum boosting ......................................................................... 118
4.3.13.4.6 drc_nr_boost_sections - number of boost sections ........................................................................ 118
4.3.13.4.7 drc_gain_section_boost - extra boost section gain ........................................................................ 118
4.3.13.4.8 drc_lev_section_boost - extra boost section control point level .................................................... 119
4.3.13.4.9 drc_gain_max_cut - maximum cut ................................................................................................ 119
4.3.13.4.10 drc_lev_max_cut - start of maximum cutting ................................................................................ 119
4.3.13.4.11 drc_nr_cut_sections - number of cut sections ................................................................................ 119
4.3.13.4.12 drc_gain_section_cut - extra cut section gain ................................................................................ 119
4.3.13.4.13 drc_lev_section_cut - extra cut section control point level ........................................................... 119
4.3.13.4.14 drc_tc_default_flag - DRC default time constants flag ................................................................. 119
4.3.13.4.15 drc_tc_attack - time constant for attacks ....................................................................................... 119
4.3.13.4.16 drc_tc_release - time constant for release ...................................................................................... 119
4.3.13.4.17 drc_tc_attack_fast - time constant fast attacks .............................................................................. 119
4.3.13.4.18 drc_tc_release_fast - time constant for fast release ....................................................................... 120
4.3.13.4.19 drc_adaptive_smoothing_flag - adaptive smoothing flag .............................................................. 120
4.3.13.4.20 drc_attack_threshold - fast attack threshold .................................................................................. 120
4.3.13.4.21 drc_release_threshold - fast release threshold ............................................................................... 120
4.3.13.5 drc_data - DRC frame-based data ....................................................................................................... 120
4.3.13.5.1 drc_gainset_size_value - gain set data size.................................................................................... 120
4.3.13.5.2 b_more_bits - more bits flag .......................................................................................................... 120
4.3.13.5.3 drc_version - DRC version ............................................................................................................ 120
4.3.13.5.4 drc2_bits - DRC gainset extension bits ......................................................................................... 120
4.3.13.5.5 drc_reset_flag - DRC reset flag ..................................................................................................... 120
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)11
4.3.13.5.6 drc_reserved - reserved bits ........................................................................................................... 120
4.3.13.6 drc_gains - DRC gains ........................................................................................................................ 120
4.3.13.6.1 drc_gain_val - DRC gain ............................................................................................................... 120
4.3.13.6.2 drc_gain_code - DRC gain codeword ............................................................................................ 121
4.3.13.7 DRC helper elements .......................................................................................................................... 121
4.3.13.7.1 nr_drc_channels - number of DRC channels ................................................................................. 121
4.3.13.7.2 nr_drc_subframes - number of DRC subframes ............................................................................ 121
4.3.14 dialogue enhancement - DE ...................................................................................................................... 121
4.3.14.1 b_de_data_present - dialogue enhancement data present flag ............................................................. 121
4.3.14.2 de_config_flag - dialogue enhancement configuration flag ................................................................ 121
4.3.14.3 de_config - dialogue enhancement configuration ............................................................................... 122
4.3.14.3.1 de_method - dialogue enhancement method ................................................................................. 122
4.3.14.3.2 de_max_gain - maximum dialogue enhancement gain .................................................................. 122
4.3.14.3.3 de_channel_config - channel configuration ................................................................................... 122
4.3.14.4 de_data - dialogue enhancement data .................................................................................................. 122
4.3.14.4.1 de_keep_pos_flag - keep position flag .......................................................................................... 122
4.3.14.4.2 de_mix_coef1_idx, de_mix_coef2_idx - dialogue panning parameters ........................................ 122
4.3.14.4.3 de_keep_data_flag - keep data flag ............................................................................................... 123
4.3.14.4.4 de_ms_proc_flag - M/S processing flag ........................................................................................ 123
4.3.14.4.5 de_par_code - parameter code ....................................................................................................... 123
4.3.14.4.6 de_signal_contribution - contribution of the signal to the enhancement ....................................... 123
4.3.14.5 Dialogue enhancement helper elements .............................................................................................. 123
4.3.14.5.1 de_nr_bands - number of parameter bands .................................................................................... 123
4.3.14.5.2 de_par[ ][ ] - dialogue enhancement parameter set ........................................................................ 123
4.3.14.5.3 de_par_prev[ ][ ] - previous dialogue enhancement parameter set ................................................ 124
4.3.14.5.4 de_abs_huffman(table_idx, code) - absolute parameter Huffman decoding ................................. 124
4.3.14.5.5 de_diff_huffman(table_idx, code) - differential parameter Huffman decoding ............................ 124
4.3.15 Extensible metadata delivery format - EMDF .......................................................................................... 124
4.3.15.0 Introduction ......................................................................................................................................... 124
4.3.15.1 emdf_payloads_substream - EMDF payloads substream .................................................................... 124
4.3.15.1.1 emdf_payload_id - EMDF payload identification ......................................................................... 124
4.3.15.1.2 emdf_payload_size - size of EMDF payload ................................................................................. 125
4.3.15.1.3 emdf_payload_byte - EMDF payload byte .................................................................................... 125
4.3.15.2 emdf_payload_config - EMDF payload configuration ....................................................................... 125
4.3.15.2.1 b_smpoffst - payload sample offset flag ........................................................................................ 125
4.3.15.2.2 smpoffst - payload sample offset ................................................................................................... 125
4.3.15.2.3 b_duration - payload duration flag ................................................................................................ 125
4.3.15.2.4 duration - payload duration - variable_bits(11) ............................................................................. 125
4.3.15.2.5 b_groupid - payload group ID flag ................................................................................................ 125
4.3.15.2.6 groupid - payload group ID ........................................................................................................... 125
4.3.15.2.7 b_codecdata - codec specific data flag .......................................................................................... 126
4.3.15.2.8 codecdata - codec specific data...................................................................................................... 126
4.3.15.2.9 b_discard_unknown_payload - discard unknown payload during transcode flag ......................... 126
4.3.15.2.10 b_payload_frame_aligned - payload to audio data frame alignment flag ...................................... 126
4.3.15.2.11 b_create_duplicate - create duplicate payload during transcode flag ............................................ 126
4.3.15.2.12 b_remove_duplicate - remove duplicate payload during transcode flag ....................................... 126
4.3.15.2.13 priority - payload priority .............................................................................................................. 126
4.3.15.2.14 proc_allowed - processing allowed ............................................................................................... 127
4.3.15.3 emdf_protection - EMDF protection data ........................................................................................... 127
4.3.15.3.1 protection_length_primary - length of protection_bits_primary field ........................................... 127
4.3.15.3.2 protection_length_secondary - length of protection_bits_secondary field - 2 bits ........................ 127
4.3.15.3.3 protection_bits_primary - primary EMDF substream protection data - 8 to 128 bits .................... 127
4.3.15.3.4 protection_bits_secondary - secondary EMDF substream protection data bits ............................. 128
4.3.16 High sampling frequency - HSF ............................................................................................................... 128
4.3.16.1 max_sfb_hsf_ext[i] - HSF extension for number of transmitted scale factor bands for index i .......... 128
4.3.16.2 get_max_sfb_hsf(g) – get max_sfb_hsf for group g ........................................................................... 128
5 Algorithmic details ............................................................................................................................... 128
5.1 audio spectral front end .................................................................................................................................. 128
5.1.1 Introduction............................................................................................................................................... 128
5.1.2 Entropy coding of spectral values ............................................................................................................. 129
5.1.2.1 Data and control interfaces .................................................................................................................. 129
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)12
5.1.2.2 Decoding process ................................................................................................................................ 129
5.1.3 Quantization reconstruction and scaling ................................................................................................... 130
5.1.3.1 Data and control interfaces .................................................................................................................. 130
5.1.3.2 Decoding process ................................................................................................................................ 131
5.1.4 Spectral noise fill ...................................................................................................................................... 132
5.1.4.1 Data and control interfaces .................................................................................................................. 132
5.1.4.2 Decoding process ................................................................................................................................ 133
5.1.5 Spectral ungrouping tool ........................................................................................................................... 134
5.1.5.1 Data and control interfaces .................................................................................................................. 134
5.1.5.2 Decoding process ................................................................................................................................ 134
5.2 Speech spectral frontend (SSF) ...................................................................................................................... 135
5.2.1 Introduction............................................................................................................................................... 135
5.2.2 Top level structure of the SSF .................................................................................................................. 136
5.2.3 Envelope decoder ...................................................................................................................................... 136
5.2.3.0 Data and control interfaces .................................................................................................................. 136
5.2.3.0a Envelope decoding .............................................................................................................................. 137
5.2.3.0b Envelope interpolation ........................................................................................................................ 137
5.2.3.0c Gain decoding ..................................................................................................................................... 138
5.2.3.0d Envelope refinement ........................................................................................................................... 138
5.2.4 Predictor decoder ...................................................................................................................................... 138
5.2.4.0 Data and control interfaces .................................................................................................................. 138
5.2.4.0a Parameter calculation .......................................................................................................................... 139
5.2.5 Spectrum decoder ..................................................................................................................................... 139
5.2.5.1 Spectrum decoder: Data and control interfaces ................................................................................... 139
5.2.5.2 Spectrum decoding process ................................................................................................................. 140
5.2.5.2.0 Introduction ................................................................................................................................... 140
5.2.5.2.1 Helper variable calculation ............................................................................................................ 140
5.2.5.2.2 Heuristic scaling and envelope allocation modification ................................................................ 140
5.2.5.2.3 Lossless decoding .......................................................................................................................... 143
5.2.5.2.4 Inverse quantisation ....................................................................................................................... 143
5.2.5.2.5 Application of heuristic inverse scaling ........................................................................................ 145
5.2.6 Subband predictor ..................................................................................................................................... 145
5.2.7 Inverse flattening ...................................................................................................................................... 148
5.2.8 Parameterization ....................................................................................................................................... 149
5.2.8.1 C matrix .............................................................................................................................................. 149
5.2.8.2 Arithmetic coding ............................................................................................................................... 150
5.2.8.3 Dither and random noise ..................................................................................................................... 156
5.3 Stereo and multichannel processing ............................................................................................................... 158
5.3.1 Introduction............................................................................................................................................... 158
5.3.2 Parameter extraction ................................................................................................................................. 159
5.3.3 Processing the channel data elements ....................................................................................................... 160
5.3.3.0 Introduction ......................................................................................................................................... 160
5.3.3.1 Processing tracks of the mono_data element ...................................................................................... 160
5.3.3.2 Processing tracks of the two_channel_data or stereo_data element .................................................... 160
5.3.3.3 Processing tracks of the three_channel_data element ......................................................................... 161
5.3.3.4 Processing tracks of the four_channel_data element ........................................................................... 161
5.3.3.5 Processing tracks of the five_channel_data element ........................................................................... 162
5.3.4 Processing the channel elements ............................................................................................................... 164
5.3.4.0 Introduction ......................................................................................................................................... 164
5.3.4.1 Processing tracks of a channel_pair_element ...................................................................................... 164
5.3.4.2 Processing tracks of a 3_0_channel_element ...................................................................................... 164
5.3.4.3 Processing tracks of a 5_X_channel_element ..................................................................................... 165
5.3.4.3.0 Introduction ................................................................................................................................... 165
5.3.4.3.1 5_X_codec_mode ∈ {SIMPLE, ASPX} ....................................................................................... 165
5.3.4.3.2 5_X_codec_mode ∈ {ASPX_ACPL_1,ASPX_ACPL_2} ............................................................ 165
5.3.4.3.3 5_X_codec_mode = ASPX_ACPL_3............................................................................................ 166
5.3.4.4 Processing tracks of the 7_X_channel_element .................................................................................. 166
5.3.4.4.0 Introduction ................................................................................................................................... 166
5.3.4.4.1 7_X_codec_mode ∈ {SIMPLE, ASPX} ....................................................................................... 166
5.3.4.4.2 7_X_codec_mode = ASPX_ACPL_1............................................................................................ 167
5.3.4.4.3 7_X_codec_mode = ASPX_ACPL_2............................................................................................ 167
5.4 96 and 192 kHz decoding ............................................................................................................................... 168
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)13
5.5 IMDCT equations and block switching .......................................................................................................... 168
5.5.1 Introduction............................................................................................................................................... 168
5.5.2 Transforms ................................................................................................................................................ 169
5.5.2.1 Data and control interfaces .................................................................................................................. 169
5.5.2.2 Decoding process ................................................................................................................................ 169
5.5.3 Block switching ........................................................................................................................................ 172
5.6 Frame alignment ............................................................................................................................................. 174
5.6.1 Introduction............................................................................................................................................... 174
5.6.2 Decoding process ...................................................................................................................................... 174
5.7 QMF domain processing ................................................................................................................................ 175
5.7.1 Introduction............................................................................................................................................... 175
5.7.2 QMF control data alignment ..................................................................................................................... 175
5.7.3 QMF analysis filterbank ........................................................................................................................... 176
5.7.3.1 Data and control interfaces .................................................................................................................. 176
5.7.3.2 Decoding process ................................................................................................................................ 176
5.7.4 QMF synthesis filterbank .......................................................................................................................... 177
5.7.4.1 Data and control interfaces .................................................................................................................. 177
5.7.4.2 Decoding process ................................................................................................................................ 178
5.7.5 Companding tool ...................................................................................................................................... 179
5.7.5.1 Data and control interfaces .................................................................................................................. 179
5.7.5.2 Decoding process ................................................................................................................................ 180
5.7.6 Advanced spectral extension tool - A-SPX ............................................................................................... 181
5.7.6.1 Data and control interfaces .................................................................................................................. 181
5.7.6.2 A-SPX specific variables .................................................................................................................... 183
5.7.6.3 Decoding A-SPX control data ............................................................................................................. 184
5.7.6.3.1 Subband groups ............................................................................................................................. 184
5.7.6.3.1.0 Introduction .............................................................................................................................. 184
5.7.6.3.1.1 Master subband group table ..................................................................................................... 185
5.7.6.3.1.2 Signal envelope subband group tables ..................................................................................... 186
5.7.6.3.1.3 Noise subband group table ....................................................................................................... 187
5.7.6.3.1.4 Patch subband group table ....................................................................................................... 187
5.7.6.3.1.5 Limiter subband group table .................................................................................................... 188
5.7.6.3.2 Low band filter and QMF delay line ............................................................................................. 189
5.7.6.3.3 Time/frequency matrix .................................................................................................................. 190
5.7.6.3.3.0 Introduction .............................................................................................................................. 190
5.7.6.3.3.1 Framing .................................................................................................................................... 190
5.7.6.3.3.2 Tiling........................................................................................................................................ 192
5.7.6.3.4 Decoding A-SPX signal and noise envelopes ................................................................................ 193
5.7.6.3.5 Dequantization and stereo decoding .............................................................................................. 195
5.7.6.4 HF signal construction ........................................................................................................................ 196
5.7.6.4.1 HF generator tool ........................................................................................................................... 196
5.7.6.4.1.1 Introduction .............................................................................................................................. 196
5.7.6.4.1.2 Pre-flattening control data calculation ..................................................................................... 196
5.7.6.4.1.3 Subband tonal to noise ratio adjustment data calculation ........................................................ 197
5.7.6.4.1.4 HF signal creation .................................................................................................................... 199
5.7.6.4.2 HF envelope adjustment tool ......................................................................................................... 200
5.7.6.4.2.0 Introduction .............................................................................................................................. 200
5.7.6.4.2.1 Estimation of transmitted and actual envelopes in the current interval .................................... 200
5.7.6.4.2.2 Calculation of compensatory gains .......................................................................................... 202
5.7.6.4.3 Noise generator tool ...................................................................................................................... 204
5.7.6.4.4 Tone generator tool ........................................................................................................................ 205
5.7.6.4.5 HF signal assembling tool ............................................................................................................. 206
5.7.6.5 Interleaved waveform coding .............................................................................................................. 207
5.7.6.5.1 Introduction ................................................................................................................................... 207
5.7.6.5.2 Signalling interleaved waveform coding ....................................................................................... 208
5.7.6.5.3 Interleaving WCC and SEC ........................................................................................................... 208
5.7.7 Advanced coupling tool - A-CPL ............................................................................................................. 209
5.7.7.1 Introduction ......................................................................................................................................... 209
5.7.7.2 Parameter band to QMF subband mapping ......................................................................................... 210
5.7.7.3 Interpolation ........................................................................................................................................ 210
5.7.7.4 Decorrelator and transient ducker ....................................................................................................... 211
5.7.7.4.1 Introduction ................................................................................................................................... 211
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)14
5.7.7.4.2 Decorrelator IIR filtering ............................................................................................................... 212
5.7.7.4.3 Transient ducker ............................................................................................................................ 212
5.7.7.5 Advanced coupling in the channel pair element .................................................................................. 214
5.7.7.6 Advanced coupling in the multichannel element ................................................................................ 215
5.7.7.6.1 5_X_codec_mode ∈ {ASPX_ACPL_1, ASPX_ACPL_2} ........................................................... 215
5.7.7.6.2 5_X_codec_mode = ASPX_ACPL_3............................................................................................ 216
5.7.7.6.3 7_X_codec_mode ∈ {ASPX_ACPL_1, ASPX_ACPL_2} ........................................................... 218
5.7.7.7 Differential decoding and dequantization ........................................................................................... 219
5.7.8 Dialogue enhancement .............................................................................................................................. 221
5.7.8.1 Introduction ......................................................................................................................................... 221
5.7.8.2 Processed channels .............................................................................................................................. 223
5.7.8.3 Dequantization .................................................................................................................................... 223
5.7.8.4 Parameter bands .................................................................................................................................. 226
5.7.8.5 Rendering ............................................................................................................................................ 226
5.7.8.6 Interpolation ........................................................................................................................................ 226
5.7.8.7 Parametric channel independent enhancement .................................................................................... 227
5.7.8.8 Parametric cross-channel enhancement ............................................................................................... 228
5.7.8.9 Waveform-parametric hybrid .............................................................................................................. 228
5.7.9 Dynamic range control tool ...................................................................................................................... 229
5.7.9.1 Introduction ......................................................................................................................................... 229
5.7.9.2 DRC Modes......................................................................................................................................... 230
5.7.9.3 Decoding process ................................................................................................................................ 230
5.7.9.3.1 Compression curves ....................................................................................................................... 230
5.7.9.3.1.0 Introduction .............................................................................................................................. 230
5.7.9.3.1.1 Level computation ................................................................................................................... 231
5.7.9.3.1.2 Gain computation ..................................................................................................................... 231
5.7.9.3.2 Directly transmitted DRC gains .................................................................................................... 232
5.7.9.3.3 Application of gain values ............................................................................................................. 233
5.7.9.4 Transcoding to a AC-3 or E-AC-3 format ........................................................................................... 233
6 Decoding the AC-4 bitstream ............................................................................................................... 233
6.1 Introduction .................................................................................................................................................... 233
6.2 Decoding process ........................................................................................................................................... 234
6.2.1 Input bitstream .......................................................................................................................................... 234
6.2.2 Structure of the bitstream .......................................................................................................................... 234
6.2.2.0 Overview ............................................................................................................................................. 234
6.2.2.1 Raw AC-4 frame ................................................................................................................................. 235
6.2.2.2 Table of contents ................................................................................................................................. 236
6.2.2.3 Presentation information ..................................................................................................................... 236
6.2.2.4 Substream ............................................................................................................................................ 236
6.2.3 Selecting and decoding a presentation ...................................................................................................... 237
6.2.4 Buffer model ............................................................................................................................................. 237
6.2.5 Decoding of a substream ........................................................................................................................... 238
6.2.5.0 Substream types .................................................................................................................................. 238
6.2.5.1 Decoding of an AC-4 substream ......................................................................................................... 238
6.2.5.2 Decoding of an AC-4 HSF extension substream ................................................................................. 238
6.2.5.3 Decoding of an EMDF payloads substream ........................................................................................ 238
6.2.6 Spectral frontend decoding ....................................................................................................................... 239
6.2.6.0 audio_data element types .................................................................................................................... 239
6.2.6.1 Mono decoding ................................................................................................................................... 239
6.2.6.2 Stereo decoding ................................................................................................................................... 239
6.2.6.3 Multichannel audio decoding .............................................................................................................. 239
6.2.6.4 Audio spectral front end ...................................................................................................................... 239
6.2.6.5 Speech spectral frontend (SSF) ........................................................................................................... 240
6.2.7 (IMDCT) and window overlap/add .......................................................................................................... 240
6.2.8 QMF analysis ............................................................................................................................................ 240
6.2.9 Companding .............................................................................................................................................. 240
6.2.10 Advanced spectral extension - A-SPX ...................................................................................................... 240
6.2.11 Advanced coupling - A-CPL .................................................................................................................... 241
6.2.12 Dialogue enhancement .............................................................................................................................. 241
6.2.13 Dynamic range control .............................................................................................................................. 242
6.2.14 QMF synthesis .......................................................................................................................................... 242
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)15
6.2.15 Sampling rate converter ............................................................................................................................ 242
6.2.16 Mixing substream outputs ......................................................................................................................... 242
6.2.16.0 Introduction ......................................................................................................................................... 242
6.2.16.1 Mixing music and effects (M+E) and dialogue ................................................................................... 243
6.2.16.2 Mixing main and associate .................................................................................................................. 243
6.2.16.3 Mixing music and effects (M+E), dialogue and associate................................................................... 244
6.2.17 Rendering a presentation .......................................................................................................................... 244
6.2.17.0 Introduction ......................................................................................................................................... 244
6.2.17.1 Generalized rendering matrix and equation ........................................................................................ 244
6.2.17.2 Downmixing from two channels into one channel .............................................................................. 244
6.2.17.3 Downmixing from three channels into two channels .......................................................................... 245
6.2.17.4 Downmixing from five channels into two channels ............................................................................ 245
6.2.17.5 Downmixing from seven channels into five channels ......................................................................... 245
6.2.17.6 Upmixing from one channel into two channels ................................................................................... 246
6.2.17.7 Upmixing from two channels into three channels ............................................................................... 246
6.2.17.8 Upmixing from three channels into five channels ............................................................................... 246
6.2.17.9 Upmixing from five channels into seven channels .............................................................................. 246
6.2.18 Decoding audio in sync with video ........................................................................................................... 246
6.2.19 Switching streams while decoding ............................................................................................................ 246
Annex A (normative): Huffman codebook tables ............................................................................ 248
A.0 Introduction .......................................................................................................................................... 248
A.1 ASF Huffman codebook tables ............................................................................................................ 248
A.2 A-SPX Huffman codebook tables ........................................................................................................ 250
A.3 A-CPL Huffman codebook tables ........................................................................................................ 253
A.4 Dialogue enhancement Huffman codebook tables ............................................................................... 257
A.5 Dynamic range control Huffman codebook table ................................................................................ 257
Annex B (normative): ASF scale factor band tables ....................................................................... 258
Annex C (normative): Speech spectral frontend tables .................................................................. 272
C.1 SSF bandwidths .................................................................................................................................... 272
C.2 POST_GAIN_LUT .............................................................................................................................. 272
C.3 PRED_GAIN_QUANT_TAB .............................................................................................................. 272
C.4 PRED_RFS_TABLE ............................................................................................................................ 273
C.5 PRED_RTS_TABLE............................................................................................................................ 273
C.6 Quantized prediction coefficients ......................................................................................................... 273
C.7 CDF_TABLE ....................................................................................................................................... 273
C.8 PREDICTOR_GAIN_CDF_LUT ........................................................................................................ 273
C.9 ENVELOPE_CDF_LUT ...................................................................................................................... 274
C.10 DITHER_TABLE ................................................................................................................................ 274
C.11 RANDOM_NOISE_TABLE ................................................................................................................ 274
C.12 STEP_SIZES_Q4_15 ........................................................................................................................... 274
C.13 AC_COEFF_MAX_INDEX ................................................................................................................ 274
C.14 dB conversion tables ............................................................................................................................ 274
Annex D (normative): Other tables .................................................................................................. 276
D.1 Channel names ..................................................................................................................................... 276
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)16
D.2 A-SPX noise table ................................................................................................................................ 276
D.3 QMF coefficients .................................................................................................................................. 276
Annex E (normative): AC-4 bitstream storage in the ISO base media file format ...................... 277
E.0 Introduction .......................................................................................................................................... 277
E.1 AC-4 track definition............................................................................................................................ 277
E.2 AC-4 sample definition ........................................................................................................................ 278
E.3 AC4SampleEntry Box .......................................................................................................................... 278
E.4 AC4SpecificBox ................................................................................................................................... 279
E.5 AC-4 audio tracks in fragmented isomedia files .................................................................................. 284
Annex F (normative): AC-4 Transport in MPEG-DASH .............................................................. 285
F.0 Introduction .......................................................................................................................................... 285
F.1 Media Presentation Description (MPD) ............................................................................................... 285
F.1.1 Overview ........................................................................................................................................................ 285
F.1.2 General MPD requirements relating to AC-4 ................................................................................................. 285
F.1.2.0 Introduction............................................................................................................................................... 285
F.1.2.1 Adaptation sets .......................................................................................................................................... 285
F.1.2.2 Representations ......................................................................................................................................... 286
F.1.2.3 AudioChannelConfiguration descriptor .................................................................................................... 286
F.1.2.4 Accessibility descriptor ............................................................................................................................. 286
F.1.3 MPD with associated audio services using AC-4 ........................................................................................... 287
F.1.3.0 Introduction............................................................................................................................................... 287
F.1.3.1 Role descriptor .......................................................................................................................................... 287
F.1.3.2 dependencyID ........................................................................................................................................... 287
F.1.4 Descriptors specific to AC-4 elementary streams .......................................................................................... 287
F.1.4.0 Introduction............................................................................................................................................... 287
F.1.4.1 AudioChannelConfiguration descriptor .................................................................................................... 287
F.1.5 MPD manifest file examples .......................................................................................................................... 288
F.1.5.0 Introduction............................................................................................................................................... 288
F.1.5.1 MPD for a single video component and single audio component ............................................................. 288
F.1.5.2 MPD for main and associated audio services delivered in separate files .................................................. 289
Annex G (normative): AC-4 Sync Frame ......................................................................................... 291
G.1 Introduction .......................................................................................................................................... 291
G.2 ac4_syncframe ...................................................................................................................................... 291
G.3 frame_size ............................................................................................................................................ 291
Annex H (informative): Bibliography ................................................................................................. 292
History ............................................................................................................................................................ 293
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)17
Intellectual Property Rights
Essential patents
IPRs essential or potentially essential to the present document may have been declared to ETSI. The information pertaining to these essential IPRs, if any, is publicly available for ETSI members and non-members, and can be found in ETSI SR 000 314: "Intellectual Property Rights (IPRs); Essential, or potentially Essential, IPRs notified to ETSI in respect of ETSI standards", which is available from the ETSI Secretariat. Latest updates are available on the ETSI Web server (https://ipr.etsi.org/).
Pursuant to the ETSI IPR Policy, no investigation, including IPR searches, has been carried out by ETSI. No guarantee can be given as to the existence of other IPRs not referenced in ETSI SR 000 314 (or the updates on the ETSI Web server) which are, or may be, or may become, essential to the present document.
Trademarks
The present document may include trademarks and/or tradenames which are asserted and/or registered by their owners. ETSI claims no ownership of these except for any which are indicated as being the property of ETSI, and conveys no right to use or reproduce any trademark and/or tradename. Mention of those trademarks in the present document does not constitute an endorsement by ETSI of products, services or organizations associated with those trademarks.
Foreword This Technical Specification (TS) has been produced by Joint Technical Committee (JTC) Broadcast of the European Broadcasting Union (EBU), Comité Européen de Normalisation ELECtrotechnique (CENELEC) and the European Telecommunications Standards Institute (ETSI).
The present document is part 1 of a multi-part deliverable covering the Digital Audio Compression (AC-4), as identified below:
Part 1: "Channel based coding";
Part 2: "Immersive and personalized audio".
NOTE: The EBU/ETSI JTC Broadcast was established in 1990 to co-ordinate the drafting of standards in the specific field of broadcasting and related fields. Since 1995 the JTC Broadcast became a tripartite body by including in the Memorandum of Understanding also CENELEC, which is responsible for the standardization of radio and television receivers. The EBU is a professional association of broadcasting organizations whose work includes the co-ordination of its members' activities in the technical, legal, programme-making and programme-exchange domains. The EBU has active members in about 60 countries in the European broadcasting area; its headquarters is in Geneva. European Broadcasting Union CH-1218 GRAND SACONNEX (Geneva) Switzerland Tel: +41 22 717 21 11 Fax: +41 22 717 24 81
The symbolic source code for tables referenced throughout the present document is contained in archive ts_10319001v010301p0.zip which accompanies the present document.
Modal verbs terminology In the present document "shall", "shall not", "should", "should not", "may", "may not", "need", "need not", "will", "will not ", "can" and "cannot" are to be interpreted as described in clause 3.2 of the ETSI Drafting Rules (Verbal forms for the expression of provisions).
"must" and "must not" are NOT allowed in ETSI deliverables except when used in direct citation.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)18
Introduction
Motivation
Digital entertainment has been progressing at a rapid pace during the past 20 years. During this period, digital audio and video compression technologies were designed and rolled out first in the context of analogue to digital migration context for broadcast services. Then technologies evolved in order to expand this digital world with greater quality through high-definition TV and surround sound. Later on, with the rapid development of Internet entertainment services accessible to an ever increasing number of consumers, and with increasing connection speeds, non-linear services and interactive services significantly transformed the entertainment experience for the end consumer. Multiple screen services and more recently companion screens and screen-to-screen applications brought truly personalized experiences to end consumers. The latest developments in entertainment technology are focusing more than ever on delivering experiences. The increased pressure to implement digitally efficient delivery systems and multiple device environments continues to drive a need for innovation in delivery formats. At the same time, compatibility and standardization of formats is necessary for effective audience reach.
In this context, AC-4 does not only provide the compression efficiency required for the broad variety of tomorrow's broadcast and broadband delivery environments but also system integration features to address particular challenges of modern media distribution, all with the flexibility to support future audio experiences:
Coded audio frame alignment with video framing (configurable) Aligning audio and video frames greatly simplifies audio and video timebase correction (A/V sync management) in the compressed domain for contribution and distribution applications. In addition, audio elementary stream chunking/fragmentation processes for multi-screen delivery can be precisely aligned with video elementary streams to eliminate the complexity of managing A/V sync end-to-end.
Built-in dialogue enhancement dialogue enhancement algorithms allow users to modify the dialogue level guided by information from the encoder or content creator, both with and without a clean (separate) dialogue track presented to the encoder.
Designed for adaptive streaming and advertisement insertions The bit rate and channel configuration can be switched without audible glitches.
Next step in non-destructive loudness and dynamic range management Intelligent and independent dynamic range and loudness controls act across a wide range of devices and applications (home theatre to mobile) and can be configured to align with numerous worldwide standards and/or recommendations. Level management is completely automated and leverages a new signalling framework to ensure that compliant programming from the content creator is passed without cascaded processing.
Dual-ended postprocessing Metadata driven postprocessing leverages media intelligence to optimize the experience across device types and ensures that only a single instance of each postprocessing algorithm is enabled throughout the entire chain.
Support for wide range of bit rates Support for lossy/low bit-rate audio up to high quality/lossless audio.
Encoding
The AC-4 encoder accepts pulse code modulation (PCM) audio and produces an encoded bitstream consistent with the present document. The specifics of the audio encoding process are not normative requirements of the present document. Nevertheless, the encoder is expected to produce a bitstream matching the syntax described in clause 4, which, when decoded according to clause 5 and clause 6, produces audio of sufficient quality for the intended application.
Decoding
The decoding process is basically the inverse of the encoding process. The decoder de-formats various types of data, such as the encoded spectral components, and applies the relevant decoding tools.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)19
Structure of the present document
The present document is structured as follows:
• Clause 4.2 specifies the details of the AC-4 bitstream syntax.
• Clause 4.3.0 specifies how to interpret bits to values which are used in the present document.
• Clause 5 specifies the various tools used in the AC-4 decoder.
• Clause 6 provides introduction into several topics and specifies how to build a working decoder.
NOTE: Clause 6 is an entry point for the present document.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)20
1 Scope The present document specifies a coded presentation of audio information and specifies the decoding process. The coded presentation specified herein is suitable for use in digital audio transmission and storage applications. The coded presentation may convey full bandwidth audio signals, along with a low-frequency enhancement signal, for multichannel playback. Additional presentations can be included, e.g. targeting listeners with visual or hearing disabilities. A wide range of encoded bit rates is supported by decoders implemented according to the present document, ranging from state-of-the-art compression to perceptually lossless rates. The coded presentation is designed with system features such as robust operation, video frame synchronicity, and seamless switching of presentations.
2 References
2.1 Normative references References are either specific (identified by date of publication and/or edition number or version number) or non-specific. For specific references, only the cited version applies. For non-specific references, the latest version of the referenced document (including any amendments) applies.
Referenced documents which are not found to be publicly available in the expected location might be found at https://docbox.etsi.org/Reference/.
NOTE: While any hyperlinks included in this clause were valid at the time of publication, ETSI cannot guarantee their long term validity.
The following referenced documents are necessary for the application of the present document.
[1] ISO/IEC 23009-1:2014: "Information technology -- Dynamic Adaptive Streaming over HTTP (DASH) -- Part 1: Media presentation description and segment formats".
[2] ISO/IEC 14496-12:2015: "Information technology -- Coding of audio-visual objects -- Part 12: ISO base media file format".
[3] ETSI TS 102 822-3-1 (V1.4.1) (11-2007): "Broadcast and On-line Services: Search, select, and rightful use of content on personal storage systems ("TV-Anytime"); Part 3: Metadata; Sub-part 1: Phase 1 - Metadata schemas".
[4] IETF BCP 47: "Tags for Identifying Languages".
[5] ISO 639-1: "Codes for the representation of names of languages; Part 1: Alpha-2 code".
[6] IETF RFC 6381: "The 'Codecs' and 'Profiles' Parameters for "Bucket" Media Types".
2.2 Informative references References are either specific (identified by date of publication and/or edition number or version number) or non-specific. For specific references, only the cited version applies. For non-specific references, the latest version of the referenced document (including any amendments) applies.
NOTE: While any hyperlinks included in this clause were valid at the time of publication, ETSI cannot guarantee their long term validity.
The following referenced documents are not necessary for the application of the present document but they assist the user with regard to a particular subject area.
[i.1] ETSI TS 102 366: "Digital Audio Compression (AC-3, Enhanced AC-3) Standard".
[i.2] ISO/IEC 13818-1 (2000): "Information technology - Generic coding of moving pictures and associated audio information: Systems".
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)21
[i.3] Recommendation ITU-R BS.1770: "Algorithms to measure audio programme loudness and true-peak audio level".
[i.4] Recommendation ITU-R BS.1771: "Requirements for loudness and true-peak indicating meters".
[i.5] ATSC Standard A/85: "Techniques for Establishing and Maintaining Audio Loudness for Digital Television".
[i.6] Recommendation EBU R128: "Loudness normalisation and permitted maximum level of audio signals".
[i.7] Technical Report ARIB TR-B32: "Operational Guidelines for Loudness of Digital Television Programs".
[i.8] Free TV Australia Operational Practice OP-59: "Measurement and Management of Loudness in Soundtracks for Television Broadcasting".
[i.9] Dolby Laboratories Speech Gating Reference Code and Information.
NOTE: Available at http://www.dolby.com/us/en/technologies/speech-gating-reference-code.aspx.
[i.10] Recommendation EBU Tech 3342: "Loudness Range: A Measure to supplement loudness normalization in accordance with EBU R128".
[i.11] Recommendation EBU Tech 3341: "Loudness Metering: 'EBU Mode' metering to supplement loudness normalisation in accordance with EBU R 128".
[i.12] William H. Press (et al.) (1992): "Numerical recipes in C: the art of scientific computing". Second Edition. Oxford.
[i.13] AC-4 EMDF datatype registry.
NOTE: Available at http://emdf-ra.org.
3 Definitions, symbols, abbreviations and conventions
3.1 Definitions For the purposes of the present document, the following terms and definitions apply:
AC-4 sync frame: optional bitstream container structure encapsulating AC-4 codec frames, used to provide synchronization and robustness against transmission errors
NOTE: This container structure may be used when the transmission system does not include information about the framing of AC-4.
accessibility: features that make a program available to people with disabilities
advanced coupling: parametric coding tool for AC-4 in the quadrature mirror filter (QMF) domain, used to improve coding efficiency for multichannel and stereo content
advanced spectral extension: parametric coding tool for AC-4 in the quadrature mirror filter (QMF) domain, used to efficiently code high frequency content
audio channel: data representing an audio signal designated to a dedicated speaker position
audio presentation: set of references to AC-4 substreams to be presented to the listener simultaneously
NOTE: An audio preselection in AC-4.
audio program component: set of audio elements characterized by an audio program component type and a language
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)22
audio program component type: classifier for an audio program component with regard to its content, such as music and effects, dialogue, visually impaired or hearing impaired
audio spectral front end: waveform coding tool for AC-4 in the modified discrete cosine transform (MDCT) domain that quantizes a spectrum based on a perceptual model and is tailored for general audio signals
audio track: signal representing one channel or object essence comprising multiple audio samples
average bit rate: mode of bit rate control that approximates a given bit rate when measured over longer periods of time
bin: position of the frequency coefficient in the MDCT domain, as in frequency bin number n
bitstream: sequence of bits
bitstream element: variable, array or matrix described by a series of one or more bits in an AC-4 bitstream
block: portion of a frame
block length: temporal extent of a block (for example measured in samples or QMF time slots)
channel configuration: targeted speaker layout
channel element: bitstream element containing one or more jointly encoded channels
channel mode: coded representation of a channel configuration
codec: system consisting of an encoder and decoder
codec frame: series of PCM samples or encoded audio data representing the same time interval for all channels in the configuration
NOTE: Decoders usually operate in a codec-frame-by-codec-frame mode.
codec frame length: temporal extent of a codec frame when decoded
codec mode: codec operation mode consisting of the channel mode plus the configuration of the coding tools used in AC-4
commentary: audio program component containing supplementary dialogue, such as a director's commentary to a movie scene
companding: parametric coding tool for AC-4 in the quadrature mirror filter (QMF) domain, used to control the temporal distribution of the quantization noise introduced in the modified discrete cosine transform (MDCT) domain
complete main: type of an audio presentation where all audio program components pertaining to that presentation are available from the main elementary stream
constant bit rate: mode of bit rate control that yields the exact same number of bits per each frame or time period
DASH descriptor: DASH element used for property signalling
NOTE: DASH descriptors share a common structure.
DASH element: XML element contained in the DASH media presentation description
DASH media presentation description: formalized XML-based description for a DASH media presentation for the purpose of providing a streaming service
DASH representation: collection of one or more audio program components, encapsulated in a delivery file format and associated with descriptive metadata (such as language or role)
decoder-specific information: record in ISO base media file format files, used for decoder configuration
dialogue enhancement: coding tool for the AC-4, used to enable the user to adjust the relative level of the dialogue to their preference
dialogue substream: substream that carries dialogue
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)23
dynamic range control: tool that limits the dynamic range of the output
extensible metadata delivery format: set of rules and data structures that enables robust signaling of metadata in an end-to-end process, involving a container, metadata payloads, and authentication protocols
frame rate: number of transmission frames decoded per second in realtime operation
independently decodable frame: independently decodable frame
ISO based media file format: media file format as specified in ISO/IEC 14496-12
Low-Frequency Effects (LFE): optional single channel of limited bandwidth (typically less than 120 Hz)
MSB: in a binary representation of an integer number, the bit that has the highest value
NOTE: In a two's complement notation of signed numbers, the most significant bit indicates the sign of the number.
music and effects: audio program component containing all audio except intelligible primary language dialogue
NOTE: Can be combined with languages other than the primary language to create a dubbed version of the program.
noise scale factor: average noise floor within the region in a QMF matrix
presentation configuration: composition of an audio presentation, characterized by the types of referenced substreams
pulse code modulation: digital representation of an analog signal where the amplitude of the signal is sampled at uniform intervals
QMF matrix: representation of the QMF domain as a matrix with QMF timeslot columns and QMF subband rows
QMF subband: frequency range represented by one row in a QMF matrix, carrying a subsampled signal
QMF subsample: single element of a QMF matrix
QMF time slot: time range represented by one column in a QMF matrix
Quadrature Mirror Filter: filter that converts a PCM sample stream into two sample streams of half the input sampling rate, so that the output data rate equals the input data rate
set: collection of zero or more items
signal scale factor: average energy of the signal within the region in a QMF matrix
speech spectral front end: waveform coding tool for AC-4 in the modified discrete cosine transform (MDCT) domain that quantizes a spectrum based on a source model of speech and is tailored for speech signals
substream: part of an AC-4 bitstream, contains audio data and corresponding metadata
table of contents: data record containing version, sequence number, frame rate and other data, and at least one presentation info element
tiling: method to determine the QMF subband group structure for each of the envelopes in A-SPX
transmission frame size: data size of a transmission frame in the bitstream domain
variable bit rate: encoding method that outputs data at a variable rate
window: weighting function associated with the IMDCT transform of a block
3.2 Symbols For the purposes of the present document, the following symbols apply:
X* complex conjugate of value X if X is a scalar; and conjugate transpose if X is a vector �x� ceiling operation, round x towards plus infinity
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)24
�x� floor operation, round x towards minus infinity · inner matrix multiplication
3.3 Abbreviations For the purposes of the present document, the following abbreviations apply:
ABR Average Bit Rate AC Audio Codec A-CPL Advanced Coupling ASF Audio Spectral Front end ASPX Advanced SPectral eXtension A-SPX Advanced SPectral eXtension ATSC Advanced Television Systems Committee CB Code Book CB_DIM Code Book Dimension CBR Constant Bit Rate CDF Cumulative Distribution Function CPL Coupling DASH Dynamic Adaptive Streaming over HTTP DC Direct Current DE Dialogue Enhancement DRC Dynamic Range Control DSI Decoder-Specific Information EMDF Extensible Metadata Delivery Format HF High Frequency HSF High Sampling Frequency HTTP Hypertext Transfer Protocol Iframe Independently decodable frame I-frame Independently decodable frame I-Frame Independently decodable frame IMDCT Inverse Modified Discrete Cosine Transform KBD Kaiser-Bessel Derived LFE Low-Frequency Effects LU Loudness Unit LUFS Loudness Unit Full Scale LUT Look-Up Table M&E Music and Effects M+E Music and Effects MDCT Modified Discrete Cosine Transform MPD Media Presentation Description MPEG Moving Picture Expert Group PCM Pulse Code Modulation QMF Quadrature Mirror Filter RMS Root Mean Square SAP Stereo Audio Processing SEC Spectral Extension Components SNF Spectral Noise Fill SPX SPectral eXtension SSF Speech Spectral Frontend TOC Table Of Contents VAR Variable VBR Variable Bit Rate WCC Waveform Coded Components XML Extensible Markup Language
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)25
3.4 Conventions Unless otherwise stated, the following convention regarding the notation is used:
• Constants are indicated by uppercase italic, e.g. NOISE_FLOOR_OFFSET.
• Tables are indicated as TABLE[idx].
• Functions are indicated as func(x).
• Variables are indicated by italic, e.g. variable.
• Vectors and vector components are indicated by bold lowercase names, e.g. vector or vectoridx.
• Matrices (and vectors of vectors) are indicated by bold uppercase single letter names, e.g. M or Mrow,column.
• Indices to tables, vectors and matrices are zero based. The top left element of matrix M is M0,0.
• Bitstream syntactic elements are indicated by the use of a different font, e.g. dynrng. All bitstream elements are described in clause 4.3.0.
• Normal pseudocode interpretation of flowcharts is assumed, with no rounding or truncation unless explicitly stated.
• Units of [dB2] refer to the approximation 1dB ≡ factor of√2,0� , i.e. 6dB ≡ factor of2,0.
• Fractional frame rate is written in "shorthand notation" as defined in table 1.
• Hexadecimal constants are denoted 0x....
• Binary constants are denoted 0b....
• Where several alternatives exist for a digit, X is used as placeholder.
• Table 1a specifies standard functions used throughout pseudocode sections. Functions with a single argument also apply to vectors and matrices by mapping the function element wise.
• Speaker and channel configurations are either specified as f/s/h.x or hp.x.h. The notation can be abbreviated to f/s/h or hp.x, respectively, if the number of the corresponding speakers is zero. Table 1b defines symbols f, s, h, x, and hp.
Table 1: Shorthand notation for frame rates
Fractional frame rate Shorthand 24 × 1000
1001� 23,976
30 × 10001001� 29,97
48 × 10001001� 47,952
60 × 10001001� 59,94
120 × 10001001� 119,88
Table 1a: Standard functions used in pseudocode
Function Meaning abs(arg) |arg| sqrt(arg) arg0,5 pow(arg1, arg2) arg1arg2
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)26
Table 1b: Symbols for speaker/channel configurations
Symbol Speakers f front s surround h height x LFE hp horizontal plane
4 Bitstream syntax
4.1 Semantics of syntax specification The following pseudocode within syntax boxes describes the order of arrival of information within the bitstream. This pseudocode is roughly based on C language syntax, but simplified for ease of reading. For bitstream elements that are larger than one bit, the order of the bits in the serial bitstream is either most-significant-bit-first (for numerical values), or left-bit-first (for bit-field values). Fields or elements contained in the bitstream are indicated with bold type. Syntactic elements are typographically distinguished by the use of a different font (e.g. dynrng).
Some AC-4 bitstream elements naturally form arrays. This syntax specification treats all bitstream elements individually, whether or not they would naturally be included in arrays. Arrays are thus described as multiple elements (as in companding_active[ch] as opposed to simply companding_active or companding_active[ ]), and control structures such as for loops are employed to increment the index ([ch] for channel in this example).
When the number of bits is variable and not bound by constants, it is indicated as VAR.
4.2 Syntax specification
4.2.1 raw_ac4_frame - Raw AC-4 frame
Table 2: Syntax of raw_ac4_frame()
Syntax No. of bits
raw_ac4_frame() { ac4_toc(); fill_area; .............................................................................. VAR byte_align .............................................................................. 0…7 for (i = 0; i < n_substreams; i++) { ac4_substream_data(); /* sub stream data stays byte aligned */ } fill_area; .............................................................................. VAR byte_align; ............................................................................. 0…7 }
4.2.2 variable_bits - Variable bits
Table 3: Syntax of variable_bits()
Syntax No. of bits
variable_bits(n_bits) { value = 0; do { value += read; .................................................................. n_bits if (b_read_more) { ................................................................... 1 value <<= n_bits;
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)27
Syntax No. of bits value += (1<<n_bits); } } while(b_read_more); return value; }
4.2.3 AC-4 frame information
4.2.3.1 ac4_toc - AC-4 table of contents
Table 4: Syntax of ac4_toc()
Syntax No. of bits
ac4_toc() { bitstream_version; ....................................................................... 2 if (bitstream_version == 3) { bitstream_version += variable_bits(2); } sequence_counter; ....................................................................... 10 if (b_wait_frames) { ...................................................................... 1 wait_frames; ........................................................................ 3 if (wait_frames > 0) reserved; ...................................................................... 2 } } fs_index; ................................................................................ 1 frame_rate_index; ........................................................................ 4 b_iframe_global; ......................................................................... 1 b_single_presentation; ................................................................... 1 if (b_single_presentation == 1) { n_presentations = 1; } else { if (b_more_presentations == 1) { ..................................................... 1 n_presentations = variable_bits(2) + 2; } else { n_presentations = 0; } } payload_base = 0; if (b_payload_base) { ..................................................................... 1 payload_base = payload_base_minus1 + 1; ............................................. 5 if (payload_base == 0x20) { payload_base += variable_bits(3); } } for (i = 0; i < n_presentations; i++) { ac4_presentation_info(); } substream_index_table(); byte_align; ............................................................................ 0…7 }
4.2.3.2 ac4_presentation_info - AC-4 presentation information
Table 5: Syntax of ac4_presentation_info()
Syntax No. of bits
ac4_presentation_info() { if (b_single_substream != 1) { ............................................................ 1 presentation_config .................................................................. 3 if (presentation_config == 7) { presentation_config += variable_bits(2);
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)28
Syntax No. of bits } } presentation_version(); if (b_single_substream != 1 && presentation_config == 6) { b_add_ emdf_substreams = 1; } else { mdcompat; ............................................................................ 3 if (b_belongs_to_presentation_id) { ................................................... 1 presentation_id = variable_bits(2); } frame_rate_multiply_info(); emdf_info(); if (b_single_substream == 1) { ac4_substream_info(); } else { b_hsf_ext; ....................................................................... 1 switch(presentation_config) { case 0: /* Music and Effects (M+E) + Dialog */ ac4_substream_info(); /* M+E */ if (b_hsf_ext == 1) { ac4_hsf_ext_substream_info(); /* M+E HSF */ } ac4_substream_info(); /* Dialog */ break; case 1: /* Main + DE */ ac4_substream_info(); /* Main */ if (b_hsf_ext == 1) { ac4_hsf_ext_substream_info(); /* Main HSF */ } ac4_substream_info(); /* DE */ break; case 2: /* Main + Associate */ ac4_substream_info(); /* Main */ if (b_hsf_ext == 1) { ac4_hsf_ext_substream_info(); /* Main HSF */ } ac4_substream_info(); /* Associate */ break; case 3: /* Music and Effects (M+E) + Dialog + Associate */ ac4_substream_info(); /* M+E */ if (b_hsf_ext == 1) { ac4_hsf_ext_substream_info(); /* M+E HSF */ } ac4_substream_info(); /* Dialog */ ac4_substream_info(); /* Associate */ break; case 4: /* Main + DE + Associate */ ac4_substream_info(); /* Main */ if (b_hsf_ext == 1) { ac4_hsf_ext_substream_info(); /* Main HSF */ } ac4_substream_info(); /* DE */ ac4_substream_info(); /* Associate */ break; case 5: /* Main + HSF ext */ ac4_substream_info(); /* Main */ if (b_hsf_ext == 1) { ac4_hsf_ext_substream_info(); /* Main HSF */ } break; default: presentation_config_ext_info(); } } b_pre_virtualized; ................................................................... 1 b_add_ emdf_substreams; .............................................................. 1 } if (b_add_ emdf_substreams) { if (n_add_ emdf_substreams == 0) { .................................................... 2 n_add_ emdf_substreams = variable_bits(2) + 4; } for (i = 0; i < n_add_ emdf_substreams; i++) { emdf_info(); } } }
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)29
4.2.3.3 presentation_version - Presentation version information
Table 6: Syntax of presentation_version()
Syntax No. of bits
presentation_version() { val = 0; while (b_tmp == 1) { ...................................................................... 1 val++; } return val; }
4.2.3.4 frame_rate_multiply_info - Frame rate multiplier information
Table 7: Syntax of frame_rate_multiply_info()
Syntax No. of bits
frame_rate_multiply_info(frame_rate_index) { switch (frame_rate_index) { case 2: case 3: case 4: if (b_multiplier) { .................................................................. 1 multiplier_bit; .................................................................. 1 } break; case 0: case 1: case 7: case 8: case 9: b_multiplier; ........................................................................ 1 break; default: break; } }
4.2.3.5 emdf_info - EMDF information
Table 8: Syntax of emdf_info()
Syntax No. of bits
emdf_info() { emdf_version; ............................................................................ 2 if ( emdf_version == 3) { emdf_version += variable_bits(2); } key_id; .................................................................................. 3 if (key_id == 7) { key_id += variable_bits(3); } if (b_ emdf_payloads_substream_info) { .................................................... 1 emdf_payloads_substream_info(); } emdf_protection(); }
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)30
4.2.3.6 ac4_substream_info - AC-4 substream information
Table 9: Syntax of ac4_substream_info()
Syntax No. of bits
ac4_substream_info() { channel_mode; ...................................................................... 1/2/4/7 if (channel_mode == 0b1111111) { channel_mode += variable_bits(2); } if (fs_index == 1) { /* 48 kHz */ if (b_sf_multiplier) { ................................................................ 1 sf_multiplier; ................................................................... 1 } } if (b_bitrate_info) { ..................................................................... 1 bitrate_indicator; ................................................................. 3/5 } if (channel_mode == 0b1111010 || channel_mode == 0b1111011 || channel_mode == 0b1111100 || channel_mode == 0b1111101) { add_ch_base; ......................................................................... 1 } if (b_content_type) { ..................................................................... 1 content_type(); } for (i = 0; i < frame_rate_factor; i++) { b_iframe; ............................................................................ 1 } if (substream_index == 3) { ............................................................... 2 substream_index += variable_bits(2); } }
4.2.3.7 content_type - Content type
Table 10: Syntax of content_type()
Syntax No. of bits
content_type() { content_classifier; ...................................................................... 3 if (b_language_indicator) { ............................................................... 1 if (b_serialized_language_tag) { ...................................................... 1 b_start_tag; ................................................................... 1 language_tag_chunk; ........................................................... 16 } else { n_language_tag_bytes; ........................................................... 6 for (i = 0; i < n_language_tag_bytes; i++) { language_tag_bytes; ....................................................... 8 } } } }
4.2.3.8 presentation_config_ext_info - Presentation configuration extended information
Table 11: Syntax of presentation_config_ext_info()
Syntax No. of bits
presentation_config_ext_info() { n_skip_bytes; ............................................................................ 5 if (b_more_skip_bytes) { .................................................................. 1 n_skip_bytes += variable_bits(2) << 5;
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)31
Syntax No. of bits } for (i = 0; i < n_skip_bytes; i++) { reserved; ............................................................................ 8 } }
4.2.3.9 ac4_hsf_ext_substream_info - AC-4 HSF extension substream information
Table 12: Syntax of ac4_hsf_ext_substream_info()
Syntax No. of bits
ac4_hsf_ext_substream_info() { if (substream_index == 3) { ............................................................... 2 substream_index += variable_bits(2); } }
4.2.3.10 emdf_payloads_substream_info - EMDF payloads substream information
Table 13: Syntax of emdf_payloads_substream_info()
Syntax No. of bits
emdf_payloads_substream_info() { if (substream_index == 3) { ............................................................... 2 substream_index += variable_bits(2); } }
4.2.3.11 substream_index_table - Substream index table
Table 14: Syntax of substream_index_table()
Syntax No. of bits
substream_index_table() { if (n_substreams == 0) { .................................................................. 2 n_substreams = variable_bits(2) + 4; } if (n_substreams == 1) { b_size_present; ..................................................................... 1 } else { b_size_present = 1; } if (b_size_present) { for (s = 0; s < n_substreams; s++) { b_more_bits; ................................................................... 1 substream_size[s]; ............................................................ 10 if (b_more_bits) { substream_size[s] += (variable_bits(2) << 10); } } } }
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)32
4.2.4 AC-4 substreams
4.2.4.1 AC-4 substreams
The ac4_substream_data() element for a specific substream index depends on the type of info element which refers to this specific substream. The mapping for the info elements defined in the present document is given in table 15.
Table 15: ac4_substream_data mapping
info element type referencing the substream ac4_substream_data element ac4_substream_info() ac4_substream() ac4_hsf_ext_substream_info() ac4_hsf_ext_substream() emdf_payloads_substream_info() emdf_payloads_substream()
4.2.4.2 ac4_substream - AC-4 substream
Table 16: Syntax of ac4_substream()
Syntax No. of bits
ac4_substream() { audio_size = audio_size_value; .......................................................... 15 if (b_more_bits) { ........................................................................ 1 audio_size += variable_bits(7) << 15; } byte_align ........................................................................... 0...7 audio_data(channel_mode, b_iframe); fill_bits; ............................................................................. VAR byte_align; ............................................................................ 0…7 metadata(b_iframe); byte_align; ............................................................................ 0…7 }
4.2.4.3 ac4_hsf_ext_substream - AC-4 high sampling frequency extension substream
In case a substream is coded in 96/192 kHz, this additional substream holds the scale factor and spectral data beyond 24 kHz. A 48 kHz decoder may skip this substream.
Table 17: Syntax of ac4_hsf_ext_substream()
Syntax No. of bits
ac4_hsf_ext_substream() { max_sfb_ext_hsf[0]; ...................................................................... 6 if (b_different_framing) { max_sfb_ext_hsf[1]; .................................................................. 6 } for (ch = 0; ch < num_channels; ch++) { sf_hsf_data(); } byte_align; ............................................................................ 0…7 }
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)33
4.2.4.4 emdf_payloads_substream - EMDF payloads substream
Table 18: Syntax of emdf_payloads_substream()
Syntax No. of bits
emdf_payloads_substream() { while (emdf_payload_id != 0) { ............................................................ 5 if ( emdf_payload_id == 31) { emdf_payload_id += variable_bits(5); } emdf_payload_config(); emdf_payload_size = variable_bits(8); for (i = 0; i < emdf_payload_size; i++) { emdf_payload_byte[i]; ........................................................... 8 } } byte_align; ............................................................................ 0…7 }
4.2.5 audio_data - Audio data
Table 19: Syntax of audio_data()
Syntax No. of bits
audio_data(channel_mode, b_iframe) { switch (channel_mode) { case mono: single_channel_element(b_iframe); break; case stereo: channel_pair_element(b_iframe); break; case 3.0: 3_0_channel_element(b_iframe); break; case 5.0: 5_X_channel_element(0, b_iframe); break; case 5.1: 5_X_channel_element(1, b_iframe); break; case 7.0: case 7.1: 7_X_channel_element(channel_mode, b_iframe); break; } }
4.2.6 Channel elements
4.2.6.1 single_channel_element - Single channel element
Table 20: Syntax of single_channel_element()
Syntax No. of bits
single_channel_element(b_iframe) { mono_codec_mode; ......................................................................... 1 if (b_iframe) { if (mono_codec_mode == ASPX) { aspx_config(); } }
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)34
Syntax No. of bits if (mono_codec_mode == SIMPLE) { mono_data(0); } else { companding_control(1); mono_data(0); aspx_data_1ch(); } }
4.2.6.2 mono_data - Mono data
Table 21: Syntax of mono_data()
Syntax No. of bits
mono_data(b_lfe) { if (b_lfe) { spec_frontend = ASF; sf_info_lfe(); } else { spec_frontend; ...................................................................... 1 sf_info(spec_frontend, 0, 0); } sf_data(spec_frontend); }
4.2.6.3 channel_pair_element - Channel pair element
Table 22: Syntax of channel_pair_element()
Syntax No. of bits
channel_pair_element(b_iframe) { stereo_codec_mode; ....................................................................... 2 if (b_iframe) { if (stereo_codec_mode == ASPX) { aspx_config(); } if (stereo_codec_mode == ASPX_ACPL_1) { aspx_config(); acpl_config_1ch(PARTIAL); } if (stereo_codec_mode == ASPX_ACPL_2) { aspx_config(); acpl_config_1ch(FULL); } } switch (stereo_codec_mode) { case SIMPLE: stereo_data(); break; case ASPX: companding_control(2); stereo_data(); aspx_data_2ch(); break; case ASPX_ACPL_1: companding_control(1); if (b_enable_mdct_stereo_proc) { ..................................................... 1 spec_frontend_m = ASF; spec_frontend_s = ASF; sf_info(ASF, 1, 0); chparam_info(); } else {
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)35
Syntax No. of bits spec_frontend_m; ............................................................... 1 sf_info(spec_frontend_m, 0, 0); spec_frontend_s; ............................................................... 1 sf_info(spec_frontend_s, 0, 1); } sf_data(spec_frontend_m); sf_data(spec_frontend_s); aspx_data_1ch(); acpl_data_1ch(); break; case ASPX_ACPL_2: companding_control(1); spec_frontend; ...................................................................... 1 sf_info(spec_frontend, 0, 0); sf_data(spec_frontend); aspx_data_1ch(); acpl_data_1ch(); break; } }
4.2.6.4 stereo_data - Stereo data
Table 23: Syntax of stereo_data()
Syntax No. of bits
stereo_data() { if (b_enable_mdct_stereo_proc) { .......................................................... 1 spec_frontend_l = ASF; spec_frontend_r = ASF; sf_info(ASF, 0, 0); chparam_info(); } else { spec_frontend_l; .................................................................... 1 sf_info(spec_frontend_l, 0, 0); spec_frontend_r; .................................................................... 1 sf_info(spec_frontend_r, 0, 0); } sf_data(spec_frontend_l); sf_data(spec_frontend_r); }
4.2.6.5 3_0_channel_element - 3.0 channel element
Table 24: Syntax of 3_0_channel_element()
Syntax No. of bits
3_0_channel_element(b_iframe) { 3_0_codec_mode; .......................................................................... 1 if (3_0_codec_mode == ASPX) { if (b_iframe) { aspx_config(); } companding_control(3); } switch (3_0_coding_config) { .............................................................. 1 case 0: stereo_data(); mono_data(0); break; case 1: three_channel_data(); break; }
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)36
Syntax No. of bits if (3_0_codec_mode == ASPX) { aspx_data_2ch(); aspx_data_1ch(); } }
4.2.6.6 5_X_channel_element - 5.X channel element
Table 25: Syntax of 5_X_channel_element()
Syntax No. of bits
5_X_channel_element(b_has_lfe, b_iframe) { 5_X_codec_mode; .......................................................................... 3 if (b_iframe) { if (5_X_codec_mode == ASPX || 5_X_codec_mode == ASPX_ACPL_1 || 5_X_codec_mode == ASPX_ACPL_2 || 5_X_codec_mode == ASPX_ACPL_3) { aspx_config(); } if (5_X_codec_mode == ASPX_ACPL_1) { acpl_config_1ch(PARTIAL); } if (5_X_codec_mode == ASPX_ACPL_2) { acpl_config_1ch(FULL); } if (5_X_codec_mode == ASPX_ACPL_3) { acpl_config_2ch(); } } if (b_has_lfe) { mono_data(1); } switch (5_X_codec_mode) { case SIMPLE: case ASPX: if (5_X_codec_mode == ASPX) { companding_control(5); } switch (coding_config) { ............................................................. 2 case 0: 2ch_mode; ...................................................................... 1 two_channel_data(); two_channel_data(); mono_data(0); break; case 1: three_channel_data(); two_channel_data(); break; case 2: four_channel_data(); mono_data(0); break; case 3: five_channel_data(); break; } if (5_X_codec_mode == ASPX) { aspx_data_2ch(); aspx_data_2ch(); aspx_data_1ch(); } break; case ASPX_ACPL_1: case ASPX_ACPL_2: companding_control(3); if (coding_config) { ................................................................. 1 three_channel_data(); } else { two_channel_data();
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)37
Syntax No. of bits } if (5_X_codec_mode == ASPX_ACPL_1) { max_sfb_master; n_side_bits n_side_bits is derived by taking the largest signalled transform length from three_channel_data() or two_channel_data() above. This largest transform length determines the number of bits as per table (104) . chparam_info(); chparam_info(); sf_data(ASF); sf_data(ASF); } if (coding_config == 0) { mono_data(0); } aspx_data_2ch(); aspx_data_1ch(); acpl_data_1ch(); acpl_data_1ch(); break; case ASPX_ACPL_3: companding_control(2); stereo_data(); aspx_data_2ch(); acpl_data_2ch(); break; } }
NOTE: n_side_bits is derived by taking the largest signalled transform length from three_channel_data() or two_channel_data() above. This largest transform length determines the number of bits as per table table 105.
4.2.6.7 two_channel_data - Two channel data
Table 26: Syntax of two_channel_data()
Syntax No. of bits
two_channel_data() { if (b_enable_mdct_stereo_proc) { .......................................................... 1 sf_info(ASF, 0, 0); chparam_info(); } else { sf_info(ASF, 0, 0); sf_info(ASF, 0, 0); } sf_data(ASF); sf_data(ASF); }
4.2.6.8 three_channel_data - Three channel data
Table 27: Syntax of three_channel_data()
Syntax No. of bits
three_channel_data() { sf_info(ASF, 0, 0); three_channel_info(); sf_data(ASF); sf_data(ASF); sf_data(ASF); }
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)38
4.2.6.9 four_channel_data - Four channel data
Table 28: Syntax of four_channel_data()
Syntax No. of bits
four_channel_data() { sf_info(ASF, 0, 0); four_channel_info(); sf_data(ASF); sf_data(ASF); sf_data(ASF); sf_data(ASF); }
4.2.6.10 five_channel_data - Five channel data
Table 29: Syntax of five_channel_data()
Syntax No. of bits
five_channel_data() { sf_info(ASF, 0, 0); five_channel_info(); sf_data(ASF); sf_data(ASF); sf_data(ASF); sf_data(ASF); sf_data(ASF); }
4.2.6.11 three_channel_info - Three channel info
Table 30: Syntax of three_channel_info()
Syntax No. of bits
three_channel_info() { chel_matsel; ............................................................................. 4 chparam_info(); chparam_info(); }
4.2.6.12 four_channel_info - Four channel info
Table 31: Syntax of four_channel_info()
Syntax No. of bits
four_channel_info() { chparam_info(); chparam_info(); chparam_info(); chparam_info(); }
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)39
4.2.6.13 five_channel_info - Five channel info
Table 32: Syntax of five_channel_info()
Syntax No. of bits
five_channel_info() { chel_matsel; ............................................................................. 4 chparam_info(); chparam_info(); chparam_info(); chparam_info(); chparam_info(); }
4.2.6.14 7_X_channel_element - 7.X channel element
Table 33: Syntax of 7_X_channel_element()
Syntax No. of bits
7_X_channel_element(channel_mode, b_iframe) { 7_X_codec_mode; .......................................................................... 2 if (b_iframe) { if (7_X_codec_mode != SIMPLE) { aspx_config(); } if (7_X_codec_mode == ASPX_ACPL_1) { acpl_config_1ch(PARTIAL); } if (7_X_codec_mode == ASPX_ACPL_2) { acpl_config_1ch(FULL); } } if (channel_mode == "7.1") { mono_data(1); /* LFE */ } if (7_X_codec_mode == ASPX_ACPL_1 || 7_X_codec_mode == ASPX_ACPL_2) { companding_control(5); } switch (coding_config) { .................................................................. 2 case 0: 2ch_mode; ........................................................................... 1 two_channel_data(); two_channel_data(); break; case 1: three_channel_data(); two_channel_data(); break; case 2: four_channel_data(); break; case 3: five_channel_data(); break; } if (7_X_codec_mode == SIMPLE || 7_X_codec_mode == ASPX) { if (b_use_sap_add_ch) { .............................................................. 1 chparam_info(); chparam_info(); } two_channel_data(); /* additional channels */ } if (7_X_codec_mode == ASPX_ACPL_1) { max_sfb_master; .... n_side_bits n_side_bits is derived by taking the largest signalled transform length from the respective X_channel_data() under the 'switch (coding_config)' statement above. For coding_config == 0, this depends on which channel pair the additional channels are derived from. This largest transform length determines the number of bits as per table (104) . chparam_info(); chparam_info();
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)40
Syntax No. of bits sf_data(ASF); sf_data(ASF); } if (coding_config == 0 || coding_config == 2) { mono_data(0); } if (7_X_codec_mode != SIMPLE) { aspx_data_2ch(); aspx_data_2ch(); aspx_data_1ch(); } if (7_X_codec_mode == ASPX) { aspx_data_2ch(); } if (7_X_codec_mode == ASPX_ACPL_1 || 7_X_codec_mode == ASPX_ACPL_2) { acpl_data_1ch(); acpl_data_1ch(); } }
NOTE: n_side_bits is derived by taking the largest signalled transform length from the respective X_channel_data() under the 'switch (coding_config)' statement above. For coding_config == 0, this depends on which channel pair the additional channels are derived from. This largest transform length determines the number of bits as per table table 105.
4.2.7 Spectral frontend
4.2.7.1 sf_info - Spectral frontend info
Table 34: Syntax of sf_info()
Syntax No. of bits
sf_info(spec_frontend, b_dual_maxsfb, b_side_limited) { if (spec_frontend == ASF) { asf_transform_info(); asf_psy_info(b_dual_maxsfb, b_side_limited); } }
4.2.7.2 sf_info_lfe - Spectral frontend info for LFE
Table 35: Syntax of sf_info_lfe()
Syntax No. of bits
sf_info_lfe() { b_long_frame = 1; /* transform length = frame_length */ max_sfb[0]; .................................................................... n_msfbl_bits num_window_groups = 1; }
NOTE: n_msfbl_bits is defined in clause 4.3.6.2.1.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)41
4.2.7.3 sf_data - Spectral frontend data
Table 36: Syntax of sf_data()
Syntax No. of bits
sf_data(spec_frontend) { if (spec_frontend == ASF) { asf_section_data(); asf_spectral_data(); asf_scalefac_data(); asf_snf_data(); } else { ssf_data(b_iframe); } }
4.2.7.4 sf_hsf_data - Spectral frontend HSF extension data
Table 36a: Syntax of sf_hsf_data()
Syntax No. of bits
sf_hsf_data() { asf_hsf_spectral_data(); asf_hsf_scalefac_data(); asf_hsf_snf_data(); }
4.2.8 Audio spectral frontend
4.2.8.1 asf_transform_info - ASF transform info
Table 37: Syntax of asf_transform_info()
Syntax No. of bits
asf_transform_info() { if (frame_len_base >= 1536) { b_long_frame; ........................................................................ 1 if (b_long_frame == 0) { transf_length[0]; ............................................................... 2 transf_length[1]; ............................................................... 2 } } else { transf_length; ...................................................................... 2 } }
4.2.8.2 asf_psy_info - ASF scale factor band info
Table 38: Syntax of asf_psy_info()
Syntax No. of bits
asf_psy_info(b_dual_maxsfb, b_side_limited) { b_different_framing = 0; if (frame_len_base >= 1536 && (b_long_frame == 0) && (transf_length[0] != transf_length[1])) {
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)42
Syntax No. of bits b_different_framing = 1; } if (b_side_limited) { max_sfb_side[0]; ........................................................... n_side_bits } else { max_sfb[0]; .................................................................... n_msfb_bits if (b_dual_maxsfb) { max_sfb_side[0]; ....................................................... n_msfb_bits } } if (b_different_framing) { if (b_side_limited) { max_sfb_side[1]; ....................................................... n_side_bits } else { max_sfb[1]; ................................................................ n_msfb_bits if (b_dual_maxsfb) { max_sfb_side[1]; ................................................... n_msfb_bits } } } for (i = 0; i < n_grp_bits; i++) { scale_factor_grouping[i] = scale_factor_grouping_bit; ................................ 1 } }
NOTE 1: n_msfb_bits and n_side_bits are defined in clause 4.3.6.2.1. NOTE 2: n_grp_bits is defined in clause 4.3.6.2.4.
4.2.8.3 asf_section_data - ASF section data
Table 39: Syntax of asf_section_data()
Syntax No. of bits
asf_section_data() { for (g = 0; g < num_window_groups; g++) { transf_length_g = get_transf_length(g); if (transf_length_g <= 2) { sect_esc_val = ( 1 << 3 ) - 1; // = 7 n_sect_bits = 3; } else { sect_esc_val = ( 1 << 5 ) - 1; // = 31 n_sect_bits = 5; } k = 0; i = 0; num_sec_lsf[g] = 0; max_sfb = get_max_sfb(g); while (k < max_sfb) { sect_cb[g][i]; ................................................................... 4 sect_len = 1; while (sect_len_incr == sect_esc_val) { .................................. n_sect_bits sect_len += sect_esc_val; } sect_len += sect_len_incr; sect_start[g][i] = k; sect_end[g][i] = k + sect_len; if (sect_start[g][i] < num_sfb_48(transf_length_g) && sect_end[g][i] >= num_sfb_48(transf_length_g)) { num_sec_lsf[g] = i + 1; if (sect_end[g][i] > num_sfb_48(transf_length_g)) { sect_end[g][i] = num_sfb_48(transf_length_g); i++; sect_start[g][i] = num_sfb_48(transf_length_g); sect_end[g][i] = k + sect_len; sect_cb[g][i] = sect_cb[g][i-1]; }
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)43
Syntax No. of bits } for (sfb = k; sfb < k + sect_len; sfb++) sfb_cb[g][sfb] = sect_cb[g][i]; k += sect_len; i++; } num_sec[g] = i; if (num_sec_lsf[g] == 0) num_sec_lsf[g] = num_sec[g]; } }
NOTE: num_sfb_48(transf_length_g) is the number of the scale factor bands for the respective transform length at 48 kHz sampling frequency as defined in table B.1.
4.2.8.4 asf_spectral_data - ASF spectral data
Table 40: Syntax of asf_spectral_data()
Syntax No. of bits
asf_spectral_data() { for (g = 0; g < num_window_groups; g++) { for (i = 0; i < num_sec_lsf[g]; i++) { if (sect_cb[g][i] != 0 && sect_cb[g][i] <= 11) { sect_start_line = sect_sfb_offset[g][sect_start[g][i]]; sect_end_line = sect_sfb_offset[g][sect_end[g][i]]; for (k = sect_start_line; k < sect_end_line;) { if (CB_DIM[sect_cb[g][i]] == 4) { quad_qspec_lines = huff_decode(sect_cb[g][i], asf_qspec_hcw); ..................... 1…16 quant_spec[k] = get_qline(quad_qspec_lines, 1); quant_spec[k+1] = get_qline(quad_qspec_lines, 2); quant_spec[k+2] = get_qline(quad_qspec_lines, 3); quant_spec[k+3] = get_qline(quad_qspec_lines, 4); if (UNSIGNED_CB[sect_cb[g][i]]) quad_sign_bits; ................................................ 0…4 k += 4; } else { /* (CB_DIM[sect_cb[g][i]] == 2) */ pair_qspec_lines = huff_decode(sect_cb[g][i], asf_qspec_hcw); ..................... 1…15 quant_spec[k] = get_qline(pair_spec_lines, 1); quant_spec[k+1] = get_qline(pair_spec_lines, 2); if (UNSIGNED_CB[sect_cb[g][i]]) pair_sign_bits; ................................................ 0…2 if (sect_cb[g][i] == 11) { if (quant_spec[k] == 16) quant_spec[k] = ext_decode(ext_code); ...................... 5…21 if (quant_spec[k+1] == 16) quant_spec[k+1] = ext_decode(ext_code); .................... 5…21 } k += 2; } } } } } }
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)44
4.2.8.5 asf_scalefac_data - ASF scale factor data
Table 41: Syntax of asf_scalefac_data()
Syntax No. of bits
asf_scalefac_data() { reference_scale_factor; .................................................................. 8 first_scf_found = 0; for (g = 0; g < num_window_groups; g++) { max_sfb = min(get_max_sfb(g), num_sfb_48(get_transf_length(g)); for (sfb = 0; sfb < max_sfb; sfb++) { if (sfb_cb[g][sfb] != 0) { if (max_quant_idx[g][sfb] > 0) { if (first_scf_found == 1) dpcm_sf[g][sfb] = huff_decode(ASF_HCB_SCALEFAC, asf_sf_hcw); ......................... 1…17 else first_scf_found = 1; } } } } }
NOTE: max_quant_idx[g][sfb] is the maximum of the absolute values of the quantized spectral lines for group g and scale factor band sfb.
4.2.8.6 asf_snf_data - ASF spectral noise fill data
Table 42: Syntax of asf_snf_data()
Syntax No. of bits
asf_snf_data(b_iframe) { if (b_snf_data_exists) { .................................................................. 1 for (g = 0; g < num_window_groups; g++) { transf_length_g = get_transf_length(g); max_sfb = min(get_max_sfb(g), num_sfb_48(transf_length_g); for (sfb = 0; sfb < max_sfb; sfb++) { if ((sfb_cb[g][sfb] == 0) || (max_quant_idx[g][sfb] == 0)) { dpcm_snf[g][sfb] = huff_decode(ASF_HCB_SNF, asf_snf_hcw); ............................ 3…8 } } } } }
4.2.8.7 asf_hsf_spectral_data - ASF spectral data for HSF extension
Table 42a: Syntax of asf_hsf_spectral_data()
Syntax No of bits
asf_hsf_spectral_data() { for (g = 0; g < num_window_groups; g++) { for (i = num_sec_lsf[g]; i < num_sec[g]; i++) { if (sect_cb[g][i] != 0 && sect_cb[g][i] <= 11) { sect_start_line = sect_sfb_offset[g][sect_start[g][i]]; sect_end_line = sect_sfb_offset[g][sect_end[g][i]]; for (k = sect_start_line; k < sect_end_line;) { if (CB_DIM[sect_cb[g][i]] == 4) { quad_qspec_lines = huff_decode(sect_cb[g][i], asf_qspec_hcw); ...................... 1…16 quant_spec[k] = get_qline(quad_qspec_lines, 1);
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)45
Syntax No of bits quant_spec[k+1] = get_qline(quad_qspec_lines, 2); quant_spec[k+2] = get_qline(quad_qspec_lines, 3); quant_spec[k+3] = get_qline(quad_qspec_lines, 4); if (UNSIGNED_CB[sect_cb[g][i]]) quad_sign_bits; .................................................. 0…4 k += 4; } else { /* (CB_DIM[sect_cb[g][i]] == 2) */ pair_qspec_lines = huff_decode(sect_cb[g][i], asf_qspec_hcw); ...................... 1…15 quant_spec[k] = get_qline(pair_spec_lines, 1); quant_spec[k+1] = get_qline(pair_spec_lines, 2); if (UNSIGNED_CB[sect_cb[g][i]]) pair_sign_bits; .................................................. 0…2 if (sect_cb[g][i] == 11) { if (quant_spec[k] == 16) quant_spec[k] = ext_decode(ext_code); ....................... 5…21 if (quant_spec[k+1] == 16) quant_spec[k+1] = ext_decode(ext_code); ..................... 5…21 } k += 2; } } } } } }
4.2.8.8 asf_hsf_scalefac_data - ASF scale factor data for HSF extension
Table 42b: Syntax of asf_hsf_scalefac_data()
Syntax No. of bits
asf_hsf_scalefac_data() { for (g = 0; g < num_window_groups; g++) { start_sfb = num_sfb_48(get_transf_length(g)); for (sfb = start_sfb; sfb < get_max_sfb_hsf(g); sfb++) { if (sfb_cb[g][sfb] != 0) { if (max_quant_idx[g][sfb] > 0) { if (first_scf_found == 1) dpcm_sf[g][sfb] = huff_decode(ASF_HCB_SCALEFAC, asf_sf_hcw); ......................... 1…17 else first_scf_found = 1; } } } } }
4.2.8.9 asf_hsf_snf_data - ASF spectral noise fill data for HSF extension
Table 42c: Syntax of ac4_hsf_snf_data()
Syntax No. of bits
asf_hsf_snf_data() { if (b_snf_data_exists) { for (g = 0; g < num_window_groups; g++) { start_sfb = num_sfb_48(get_transf_length(g)); for (sfb = start_sfb; sfb < get_max_sfb_hsf(g); sfb++) { if ((sfb_cb[g][sfb] == 0) || (max_quant_idx[g][sfb] == 0)) { dpcm_snf[g][sfb] = huff_decode(ASF_HCB_SNF, asf_snf_hcw); ............................ 3…8 }
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)46
Syntax No. of bits } } } }
4.2.9 Speech spectral frontend
4.2.9.1 ssf_data - Speech spectral frontend data
Table 43: Syntax of ssf_data()
Syntax No. of bits
ssf_data(b_iframe) { if (b_iframe) { b_ssf_iframe = 1; } else { b_ssf_iframe; ....................................................................... 1 } ssf_granule(b_ssf_iframe); if (frame_len_base >= 1536) { ssf_granule(0); } }
4.2.9.2 ssf_granule - Speech spectral frontend granule
Table 44: Syntax of ssf_granule()
Syntax No. of bits
ssf_granule(b_iframe) { stride_flag; ............................................................................. 1 if (b_iframe == 1) { num_bands_minus12; .................................................................. 3 num_bands = num_bands_minus12 + 12; } start_block = 0; end_block = 0; if ((stride_flag == LONG_STRIDE) && (b_iframe == 0)) { end_block = 1; } if (stride_flag == SHORT_STRIDE) { end_block = 4; if (b_iframe ==1) { start_block = 1; } } for (block = start_block; block < end_block; block++) { predictor_presence_flag[block]; ..................................................... 1 if (predictor_presence_flag[block] == 1) { if ((start_block == 1) && (block == 1)) { delta_flag[block] = 0; } else { delta_flag[block]; ........................................................ 1 } } } ssf_st_data(); ssf_ac_data(); }
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)47
4.2.9.3 ssf_st_data - Speech spectral frontend static data
Table 45: Syntax of ssf_st_data()
Syntax No. of bits
ssf_st_data() { env_curr_band0_bits; ..................................................................... 5 if ((b_iframe == 1) && (stride_flag == SHORT_STRIDE)) { env_startup_band0_bits; ............................................................. 5 } if (stride_flag == SHORT_STRIDE) { for (block = 0; block < 4; block++) { gain_bits[block]; .............................................................. 4 } } num_blocks = (stride_flag == SHORT_STRIDE) ? 4 : 1; for (block = 0; block < num_blocks; block++) { if (block >= start_block && block < end_block) { if (predictor_presence_flag[block] == 1) { if (delta_flag[block] == 1) { predictor_lag_delta_bits[block]; ....................................... 4 } else { predictor_lag_bits[block]; ............................................ 9 } } } variance_preserving_flag[block]; .................................................... 1 alloc_offset_bits[block]; ........................................................... 5 } }
4.2.9.4 ssf_ac_data - Speech spectral frontend arithmetic coded data
Table 46: Syntax of ssf_ac_data()
Syntax No. of bits
ssf_ac_data() { env_curr_ac_bits; // num_bands-1 values ............ VAR if ((b_iframe == 1) && (stride_flag == SHORT_STRIDE)) { env_startup_ac_bits; // num_bands-1 values .............. VAR } num_blocks = (stride_flag == SHORT_STRIDE) ? 4 : 1; for (block = 0; block < num_blocks; block++) { if (((b_iframe == 1) && (block > 0)) || (b_iframe == 0)) { if (predictor_presence_flag[block] == 1) { predictor_gain_ac_bits[block]; // 1 value ........................ VAR } } q_mdct_coefficients_ac_bits[block]; // num_bins values ................. VAR } }
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)48
4.2.10 Stereo audio processing
4.2.10.1 chparam_info -Stereo information
Table 47: Syntax of chparam_info()
Syntax No. of bits
chparam_info() { sap_mode; ................................................................................. 2 if (sap_mode == 1) { for (g = 0; g < num_window_groups; g++) { max_sfb_g = get_max_sfb(g); for (sfb = 0; sfb < max_sfb_g; sfb++) { ms_used[g][sfb]; .............................................................. 1 } } } if (sap_mode == 3) { sap_data(); } }
4.2.10.2 sap_data - Stereo audio processing data
Table 48: Syntax of sap_data()
Syntax No. of bits
sap_data() { sap_coeff_all; ........................................................................... 1 if (sap_coeff_all == 0) { for (g = 0; g < num_window_groups; g++) { max_sfb_g = get_max_sfb(g); for (sfb = 0; sfb < max_sfb_g; sfb += 2) { sap_coeff_used[g][sfb]; ...................................................... 1 if ((sfb+1) < max_sfb_g) { sap_coeff_used[g][sfb+1] = sap_coeff_used[g][sfb]; } } } } else { for (g = 0; g < num_window_groups; g++) { max_sfb_g = get_max_sfb(g); for (sfb = 0; sfb < max_sfb_g; sfb++) { sap_coeff_used[g][sfb] = 1; } } } if (num_window_groups != 1) { delta_code_time; ..................................................................... 1 } for (g = 0; g < num_window_groups; g++) { max_sfb_g = get_max_sfb(g); for (sfb = 0; sfb < max_sfb_g; sfb += 2) { if (sap_coeff_used[g][sfb]) { dpcm_alpha_q[g][sfb] = huff_decode( ASF_HCB_SCALEFAC, sap_hcw); .............................. 1…17 } } } }
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)49
4.2.11 Companding control
Table 49: Syntax of companding_control()
Syntax No. of bits
companding_control(num_chan) { sync_flag = 0; if (num _chan > 1) { sync_flag; .......................................................................... 1 } b_need_avg = 0; nc = sync_flag ? 1 : num_chan; for (ch = 0; ch < nc; ch++) { b_compand_on[ch]; ................................................................... 1 if (!b_compand_on[ch]) { b_need_avg = 1; } } if (b_need_avg) { b_compand_avg; ...................................................................... 1 } }
4.2.12 Advanced spectral extension - A-SPX
4.2.12.1 aspx_config - A-SPX configuration
Table 50: Syntax of aspx_config()
Syntax No. of bits
aspx_config() { aspx_quant_mode_env; ..................................................................... 1 aspx_start_freq; ......................................................................... 3 aspx_stop_freq; .......................................................................... 2 aspx_master_freq_scale; .................................................................. 1 aspx_interpolation; ...................................................................... 1 aspx_preflat; ............................................................................ 1 aspx_limiter; ............................................................................ 1 aspx_noise_sbg; .......................................................................... 2 aspx_num_env_bits_fixfix; ................................................................ 1 aspx_freq_res_mode; ...................................................................... 2 }
4.2.12.2 aspx_data_1ch - A-SPX 1-channel data
Table 51: Syntax of aspx_data_1ch()
Syntax No. of bits
aspx_data_1ch(b_iframe) { if (b_iframe) aspx_xover_subband_offset; .......................................................... 3 aspx_framing(0); aspx_qmode_env[0] = aspx_quant_mode_env; if (aspx_int_class[0] == FIXFIX && aspx_num_env[0] == 1) aspx_qmode_env[0] = 0; aspx_delta_dir(0); aspx_hfgen_iwc_1ch(); aspx_data_sig[0] = aspx_ec_data(SIGNAL, aspx_num_env[0], aspx_freq_res[0], aspx_qmode_env[0], 0,
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)50
Syntax No. of bits aspx_sig_delta_dir[0]); aspx_data_noise[0] = aspx_ec_data( NOISE, aspx_num_noise[0], 0, 0, 0, aspx_noise_delta_dir[0]); }
4.2.12.3 aspx_data_2ch - A-SPX 2-channel data
Table 52: Syntax of aspx_data_2ch()
Syntax No. of bits
aspx_data_2ch(b_iframe) { if (b_iframe) aspx_xover_subband_offset; ........................................................... 3 aspx_framing(0); aspx_qmode_env[0] = aspx_qmode_env[1] = aspx_quant_mode_env; if (aspx_int_class[0] == FIXFIX && aspx_num_env[0] == 1) aspx_qmode_env[0] = aspx_qmode_env[1] = 0; aspx_balance; ............................................................................ 1 if (aspx_balance == 0) { aspx_framing(1); aspx_qmode_env[1] = aspx_quant_mode_env; if (aspx_int_class[1] == FIXFIX && aspx_num_env[1] == 1) aspx_qmode_env[1] = 0; } aspx_delta_dir(0); aspx_delta_dir(1); aspx_hfgen_iwc_2ch(aspx_balance); aspx_data_sig[0] = aspx_ec_data(SIGNAL, aspx_num_env[0], aspx_freq_res[0], aspx_qmode_env[0], LEVEL, aspx_sig_delta_dir[0]); aspx_data_sig[1] = aspx_ec_data(SIGNAL, aspx_num_env[1], aspx_freq_res[1], aspx_qmode_env[1], aspx_balance ? BALANCE : LEVEL, aspx_sig_delta_dir[1]); aspx_data_noise[0] = aspx_ec_data(NOISE, aspx_num_noise[0], 0, 0, LEVEL, aspx_noise_delta_dir[0]); aspx_data_noise[1] = aspx_ec_data(NOISE, aspx_num_noise[1], 0, 0, aspx_balance ? BALANCE : LEVEL, aspx_noise_delta_dir[1]); }
4.2.12.4 aspx_framing - A-SPX framing
Table 53: Syntax of aspx_framing()
Syntax No. of bits
aspx_framing(ch) { aspx_num_rel_left[ch] = 0; aspx_num_rel_right[ch] = 0;
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)51
Syntax No. of bits aspx_int_class[ch]; .................................................................. 1...3 switch (aspx_int_class[ch]) { case FIXFIX: envbits = aspx_num_env_bits_fixfix + 1; aspx_num_env[ch] = (1 << tmp_num_env); ......................................... envbits if (aspx_freq_res_mode == 0) aspx_freq_res[ch][0]; ............................................................ 1 break; case FIXVAR: aspx_var_bord_right[ch]; ............................................................. 2 aspx_num_rel_right[ch]; // Note 1 ................................................. 2 (1) for (rel = 0; rel < aspx_num_rel_right[ch]; rel++) aspx_rel_bord_right[ch][rel] = 2*tmp + 2; // Note 1 ................. 2 (1) break; case VARVAR: if (b_iframe) aspx_var_bord_left[ch]; .......................................................... 2 aspx_num_rel_left[ch]; // Note 1 ................. 2 (1) for (rel = 0; rel < aspx_num_rel_left[ch]; rel++) aspx_rel_bord_left[ch][rel] = 2*tmp + 2; // Note 1 ................. 2 (1) aspx_var_bord_right[ch]; ............................................................. 2 aspx_num_rel_right[ch]; // Note 2 ................. 2 (1) for (rel = 0; rel < aspx_num_rel_right[ch]; rel++) aspx_rel_bord_right[ch][rel] = 2*tmp + 2; // Note 1 ................. 2 (1) break; case VARFIX: if (b_iframe) aspx_var_bord_left[ch]; .......................................................... 2 aspx_num_rel_left[ch]; // Note 1 ................. 2 (1) for (rel = 0; rel < aspx_num_rel_left[ch]; rel++) aspx_rel_bord_left[ch][rel] = 2*tmp + 2; // Note 1 ................. 2 (1) break; } if (aspx_int_class[ch] != FIXFIX) { aspx_num_env[ch]=aspx_num_rel_left[ch]+aspx_num_rel_right[ch]+1; ptr_bits = ceil(log(aspx_num_env[ch]+2) / log(2)); // Note 2 aspx_tsg_ptr[ch] = tmp - 1; .................................................... ptr_bits if (aspx_freq_res_mode == 0) for (env = 0; env < aspx_num_env[ch]; env++) aspx_freq_res[ch][env]; ...................................................... 1 } if (aspx_num_env[ch] > 1) aspx_num_noise[ch] = 2; else aspx_num_noise[ch] = 1; }
NOTE 1: The value within parenthesis applies when num_aspx_timeslots ≤ 8. NOTE 2: The division (/) is a float division without rounding or truncation.
4.2.12.5 aspx_delta_dir - A-SPX direction of envelope delta coding
Table 54: Syntax of aspx_delta_dir()
Syntax No. of bits
aspx_delta_dir(ch) { for (env = 0; env < aspx_num_env[ch]; env++) // Note aspx_sig_delta_dir[ch][env]; ......................................................... 1 for (env = 0; env < aspx_num_noise[ch]; env++) // Note aspx_noise_delta_dir[ch][env]; ....................................................... 1 }
NOTE: aspx_num_env and aspx_num_noise are defined in clause 4.2.12.4.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)52
4.2.12.6 aspx_hfgen_iwc_1ch - A-SPX 1-channel HF generation and interleaved waveform coding
Table 55: Syntax of aspx_hfgen_iwc_1ch()
Syntax No. of bits
aspx_hfgen_iwc_1ch() { for (n = 0; n < num_sbg_noise; n++) // Note 1 aspx_tna_mode[n]; .................................................................... 2 if (aspx_ah_present) { .................................................................... 1 for (n = 0; n < num_sbg_sig_highres; n++) // Note 1 aspx_add_harmonic[n]; ............................................................ 1 } /* initialize frequency interleaved coding flags to zero */ for (n = 0; n < num_sbg_sig_highres; n++) // Note 1 aspx_fic_used_in_sfb[n] = 0; /* read frequency interleaved coding flags */ if (aspx_fic_present) { ................................................................... 1 for (n = 0; n < num_sbg_sig_highres; n++) { // Note 1 aspx_fic_used_in_sfb[n]; ......................................................... 1 } } /* initialize time interleaved coding flags to zero */ for (n = 0; n < num_aspx_timeslots; n++) { // Note 2 aspx_tic_used_in_slot[n] = 0; } /* read time interleaved coding flags */ if (aspx_tic_present) { ................................................................... 1 for (n = 0; n < num_aspx_timeslots; n++) { // Note 2 aspx_tic_used_in_slot[n]; ........................................................ 1 } } }
NOTE 1: Variables num_sbg_sig_highres and num_sbg_noise are derived according to clause 5.7.6.3.1. NOTE 2: Variable num_aspx_timeslots is derived in clause 5.7.6.3.3.
4.2.12.7 aspx_hfgen_iwc_2ch - A-SPX 2-channel HF generation and interleaved waveform coding
Table 56: Syntax of aspx_hfgen_iwc_2ch()
Syntax No. of bits
aspx_hfgen_iwc_2ch(aspx_balance) { for (n = 0; n < num_sbg_noise; n++) // Note 1 aspx_tna_mode[0][n]; ................................................................. 2 if (aspx_balance == 0) { for (n = 0; n < num_sbg_noise; n++) // Note 1 aspx_tna_mode[1][n]; ............................................................. 2 } else { for (n = 0; n < num_sbg_noise; n++) // Note 1 aspx_tna_mode[1][n] = aspx_tna_mode[0][n] } if (aspx_ah_left) { ....................................................................... 1 for (n = 0; n < num_sbg_sig_highres; n++) // Note 1 aspx_add_harmonic[0][n]; ......................................................... 1 } if (aspx_ah_right) { ...................................................................... 1 for (n = 0; n < num_sbg_sig_highres; n++) // Note 1 aspx_add_harmonic[1][n]; ......................................................... 1 } /* initialize frequency interleaved coding flags to zero */ for (n = 0; n < num_sbg_sig_highres; n++) // Note 1 aspx_fic_used_in_sfb[0][n] = aspx_fic_used_in_sfb[1][n] = 0; /* read frequency interleaved coding flags */ if (aspx_fic_present) { ................................................................... 1 if (aspx_fic_left) { .................................................................. 1 for (n = 0; n < num_sbg_sig_highres; n++) // Note 1 aspx_fic_used_in_sfb[0][n]; .................................................. 1
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)53
Syntax No. of bits } if (aspx_fic_right) { ................................................................. 1 for (n = 0; n < num_sbg_sig_highres; n++){ // Note 1 aspx_fic_used_in_sfb[1][n]; .................................................. 1 } } /* initialize time interleaved coding flags to zero */ for (n = 0; n < num_aspx_timeslots; n++) aspx_tic_used_in_slot[0][n] = aspx_tic_used_in_slot[1][n] = 0; aspx_tic_left = aspx_tic_right = 0; /* read time interleaved coding flags */ if (aspx_tic_present) { ................................................................... 1 aspx_tic_copy; ....................................................................... 1 if (aspx_tic_copy == 0) { aspx_tic_left; ................................................................... 1 aspx_tic_right; .................................................................. 1 } if (aspx_tic_copy || aspx_tic_left) { for (n = 0; n < num_aspx_timeslots; n++) // Note 2 aspx_tic_used_in_slot[0][n]; ................................................. 1 } if (aspx_tic_right) { for (n = 0; n < num_aspx_timeslots; n++) // Note 2 aspx_tic_used_in_slot[1][n]; ................................................. 1 } if (aspx_tic_copy) { for (n = 0; n < num_aspx_timeslots; n++) // Note 2 aspx_tic_used_in_slot[1][n]=aspx_tic_used_in_slot[0][n]; } } }
NOTE 1: Variables num_sbg_sig_highres and num_sbg_noise are derived according to clause 5.7.6.3.1. NOTE 2: Variable num_aspx_timeslots is derived in clause 5.7.6.3.3.
4.2.12.8 aspx_ec_data - A-SPX entropy coded data
Table 57: Syntax of aspx_ec_data()
Syntax No. of bits
aspx_ec_data(data_type, num_env, freq_res, quant_mode, stereo_mode, direction) { for (env = 0; env < num_env; env++) { if (data_type == SIGNAL) { if (freq_res[env]) num_sbg = num_sbg_sig_highres; // Note else num_sbg = num_sbg_sig_lowres; // Note } else { num_sbg = num_sbg_noise; // Note } qm = quant_mode; sm = stereo_mode; dir = direction[env]; a_huff_data[env] = aspx_huff_data(data_type,num_sbg,qm,sm,dir); } return a_huff_data; }
NOTE: Variables num_sbg_sig_highres and num_sbg_sig_lowres are derived in clause 5.7.6.3.1.2 and num_sbg_noise is derived according to clause 5.7.6.3.1.3.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)54
4.2.12.9 aspx_huff_data - A-SPX Huffman data
Table 58: Syntax of aspx_huff_data()
Syntax No. of bits
aspx_huff_data(data_type, num_sbg, quant_mode, stereo_mode, direction) { if (direction == 0) { // FREQ aspx_hcb = get_aspx_hcb(data_type, quant_mode, stereo_mode, F0); a_huff_data[0] = huff_decode(aspx_hcb, aspx_hcw); .................................. 1…x aspx_hcb = get_aspx_hcb(data_type, quant_mode, stereo_mode, DF); for (i = 1; i < num_sbg; i++) { a_huff_data[i] = huff_decode_diff(aspx_hcb, aspx_hcw); ........................ 1…x } } else { // TIME aspx_hcb = get_aspx_hcb(data_type, quant_mode, stereo_mode, DT); for (i = 0; i < num_sbg; i++) { a_huff_data[i] = huff_decode_diff (aspx_hcb, aspx_hcw); ....................... 1…x } } return a_huff_data; }
NOTE: The function get_aspx_hcb() is defined in clause 5.7.6.3.4.
4.2.13 Advanced coupling - A-CPL
4.2.13.1 acpl_config_1ch - A-CPL 1-channel configuration
Table 59: Syntax of acpl_config_1ch()
Syntax No. of bits
acpl_config_1ch(acpl_1ch_mode) { acpl_qmf_band = 0; acpl_param_band = 0; acpl_num_param_bands_id; ................................................................. 2 acpl_quant_mode; ......................................................................... 1 if (acpl_1ch_mode == PARTIAL) { acpl_qmf_band = acpl_qmf_band_minus1 + 1; ............................................ 3 acpl_param_band = sb_to_pb(acpl_qmf_band); } }
NOTE: acpl_1ch_mode is a helper element defined in table 141. The function sb_to_pb() maps from QMF subbands to parameter bands according to table 196.
4.2.13.2 acpl_config_2ch - A-CPL 2-channel configuration
Table 60: Syntax of acpl_config_2ch()
Syntax No. of bits
acpl_config_2ch() { acpl_qmf_band = 0; acpl_param_band = 0; acpl_num_param_bands_id; ................................................................. 2 acpl_quant_mode_0; ....................................................................... 1 acpl_quant_mode_1; ....................................................................... 1 }
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)55
4.2.13.3 acpl_data_1ch - A-CPL 1-channel data
Table 61: Syntax of acpl_data_1ch()
Syntax No. of bits
acpl_data_1ch() { acpl_framing_data(); num_bands = acpl_num_param_bands_id; start = acpl_param_band; acpl_alpha1 = acpl_ec_data(ALPHA, num_bands, start, acpl_quant_mode); acpl_beta1 = acpl_ec_data(BETA, num_bands, start, acpl_quant_mode); }
4.2.13.4 acpl_data_2ch - A-CPL 2-channel data
Table 62: Syntax of acpl_data_2ch()
Syntax No. of bits
acpl_data_2ch() { acpl_framing_data(); num_bands = acpl_num_param_bands; st = acpl_param_band; acpl_alpha1 = acpl_ec_data(ALPHA, num_bands, st, acpl_quant_mode_0); acpl_alpha2 = acpl_ec_data(ALPHA, num_bands, st, acpl_quant_mode_0); acpl_beta1 = acpl_ec_data(BETA, num_bands, st, acpl_quant_mode_0); acpl_beta2 = acpl_ec_data(BETA, num_bands, st, acpl_quant_mode_0); acpl_beta3 = acpl_ec_data(BETA3, num_bands, st, acpl_quant_mode_0); acpl_gamma1 = acpl_ec_data(GAMMA, num_bands, st, acpl_quant_mode_1); acpl_gamma2 = acpl_ec_data(GAMMA, num_bands, st, acpl_quant_mode_1); acpl_gamma3 = acpl_ec_data(GAMMA, num_bands, st, acpl_quant_mode_1); acpl_gamma4 = acpl_ec_data(GAMMA, num_bands, st, acpl_quant_mode_1); acpl_gamma5 = acpl_ec_data(GAMMA, num_bands, st, acpl_quant_mode_1); acpl_gamma6 = acpl_ec_data(GAMMA, num_bands, st, acpl_quant_mode_1); }
4.2.13.5 acpl_framing_data - A-CPL framing data
Table 63: Syntax of acpl_framing_data()
Syntax No. of bits
acpl_framing_data() { acpl_interpolation_type; ................................................................. 1 acpl_num_param_sets_cod; ................................................................. 1 if (acpl_interpolation_type == 1) { for (ps = 0; ps < acpl_num_param_sets_cod + 1; ps++) { acpl_param_timeslot[ps]; ....................................................... 5 } } }
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)56
4.2.13.6 acpl_ec_data - A-CPL entropy coded data
Table 64: Syntax of acpl_ec_data()
Syntax No. of bits
acpl_ec_data(data_type, data_bands, start_band, quant_mode) { for (ps = 0; ps < acpl_num_param_sets_cod + 1; ps++) { a_param_set[ps] = acpl_huff_data( data_type, data_bands, start_band, quant_mode); } return a_param_set; }
4.2.13.7 acpl_huff_data - A-CPL Huffman data
Table 65: Syntax of acpl_huff_data()
Syntax No. of bits
acpl_huff_data(data_type, data_bands, start_band, quant_mode) { diff_type; ............................................................................... 1 if (diff_type == 0) { // DIFF_FREQ acpl_hcb = get_acpl_hcb(data_type, quant_mode, F0); a_huff_data[start_band] = huff_decode_diff(acpl_hcb, acpl_hcw); ..................... 1…x acpl_hcb = get_acpl_hcb(data_type, quant_mode, DF); for (i = start_band + 1; i < data_bands; i++) { a_huff_data[i] = huff_decode_diff(acpl_hcb, acpl_hcw); .......................... 1…x } } else { // DIFF_TIME acpl_hcb = get_acpl_hcb(data_type, quant_mode, DT); for (i = start_band ; i < data_bands; i++) { a_huff_data[i] = huff_decode_diff(acpl_hcb, acpl_hcw); .......................... 1…x } } return a_huff_data; }
NOTE: The function get_acpl_hcb() is defined in clause 4.3.11.6.1.
4.2.14 Metadata
4.2.14.1 metadata() - Metadata
Table 66: Syntax of metadata()
Syntax No. of bits
metadata(b_iframe) { basic_metadata(); extended_metadata(); tools_metadata_size = tools_metadata_size_value; ......................................... 7 if (b_more_bits) { ........................................................................ 1 tools_metadata_size += variable_bits(3) << 7; } drc_frame(b_iframe); dialog_enhancement(b_iframe); if (b_emdf_payloads_substream) { .......................................................... 1 emdf_payloads_substream(); } }
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)57
4.2.14.2 basic_metadata - Basic metadata
Table 67: Syntax of basic_metadata()
Syntax No. of bits
basic_metadata(channel_mode) { dialnorm_bits; ........................................................................... 7 if (b_more_basic_metadata) { .............................................................. 1 if (b_further_loudness_info) { ....................................................... 1 further_loudness_info(); } if (channel_mode == stereo) { if (b_prev_dmx_info) { .......................................................... 1 pre_dmixtyp_2ch; .......................................................... 3 phase90_info_2ch; ......................................................... 2 } } if (channel_mode > stereo) { if (b_dmx_coeff) { .............................................................. 1 loro_center_mixgain; ...................................................... 3 loro_surround_mixgain; .................................................... 3 if (b_loro_dmx_loud_corr) { ................................................ 1 loro_dmx_loud_corr; .................................................. 5 } if (b_ltrt_mixinfo) { ...................................................... 1 ltrt_center_mixgain; ................................................. 3 ltrt_surround_mixgain; ............................................... 3 } if (b_ltrt_dmx_loud_corr) { ................................................ 1 ltrt_dmx_loud_corr; .................................................. 5 } if (channel_mode_contains_Lfe()) { if (b_lfe_mixinfo) { .................................................. 1 lfe_mixgain; .................................................... 5 } } preferred_dmx_method; ..................................................... 2 } if (channel_mode == 5_X) { if (b_predmxtyp_5ch) { ..................................................... 1 pre_dmixtyp_5ch; ...................................................... 3 } if (b_preupmixtyp_5ch) { ................................................... 1 pre_upmixtyp_5ch; ..................................................... 4 } } if (5≤ch_mode≤10) { if (b_upmixtyp_7ch) { ...................................................... 1 if (5≤ch_mode≤6) { pre_upmixtyp_3_4; ............................................... 2 } else if (9≤ch_mode≤10) { pre_upmixtyp_3_2_2; ............................................. 1 } } } phase90_info_mc; ............................................................... 2 b_surround_attenuation_known; .................................................. 1 b_lfe_attenuation_known; ........................................................ 1 } if (b_dc_blocking) { ................................................................. 1 dc_block_on; ................................................................... 1 } } }
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)58
4.2.14.3 further_loudness_info - Additional loudness information
Table 68: Syntax of further_loudness_info()
Syntax No. of bits
further_loudness_info() { loudness_version; ........................................................................ 2 if (loudness_version == 3) { loudness_version += extended_loudness_version; ...................................... 4 } loud_prac_type; .......................................................................... 4 if (loud_prac_type != 0) { if (b_loudcorr_dialgate) { ........................................................... 1 dialgate_prac_type; ............................................................ 3 } b_loudcorr_type; .................................................................... 1 } if (b_loudrelgat) { ....................................................................... 1 loudrelgat; ........................................................................ 11 } if (b_loudspchgat) { ...................................................................... 1 loudspchgat; ....................................................................... 11 dialgate_prac_type; ................................................................. 3 } if (b_loudstrm3s) { ....................................................................... 1 loudstrm3s; ........................................................................ 11 } if (b_max_loudstrm3s) { ................................................................... 1 max_loudstrm3s; .................................................................... 11 } if (b_truepk) { ........................................................................... 1 truepk; ............................................................................ 11 } if (b_max_truepk) { ....................................................................... 1 max_truepk; ........................................................................ 11 } if (b_prgmbndy) { ......................................................................... 1 prgmbndy = 1; prgmbndy_bit = 0; while (prgmbndy_bit == 0) { prgmbndy <<= 1; prgmbndy_bit; .................................................................. 1 } b_end_or_start; ..................................................................... 1 if (b_prgmbndy_offset) { ............................................................. 1 prgmbndy_offset; .............................................................. 11 } } if (b_lra) { .............................................................................. 1 lra; ............................................................................... 10 lra_prac_type; ...................................................................... 3 } if (b_loudmntry) { ........................................................................ 1 loudmntry; ......................................................................... 11 } if (b_max_loudmntry) { .................................................................... 1 max_loudmntry; ..................................................................... 11 } if (b_extension) { ........................................................................ 1 e_bits_size; ........................................................................ 5 if (e_bits_size == 31) { e_bits_size += variable_bits(4); } extension_bits; ........................................................... e_bits_size } }
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)59
4.2.14.4 extended_metadata - Extended metadata
Table 69: Syntax of extended_metadata()
Syntax No. of bits
extended_metadata(channel_mode, b_associated, b_dialog) { if (b_associated) { if (b_scale_main) { .................................................................. 1 scale_main; .................................................................... 8 } if (b_scale_main_center) { ........................................................... 1 scale_main_center; ............................................................. 8 } if (b_scale_main_front) { ............................................................ 1 scale_main_front; .............................................................. 8 } if (channel_mode == mono) { pan_associated; ................................................................ 8 } } if (b_dialog) { if (b_dialog_max_gain) { ............................................................. 1 dialog_max_gain; ............................................................... 2 } if (b_pan_dialog_present) { .......................................................... 1 if (channel_mode == mono) { pan_dialog; ............................................................... 8 } else { pan_dialog[0]; ............................................................ 8 pan_dialog[1]; ............................................................ 8 pan_signal_selector; ..................................................... 2 } } } if (b_channels_classifier) { .............................................................. 1 if (channel_mode_contains_c()) { if (b_c_active) { ............................................................... 1 b_c_has_dialog; ............................................................. 1 } } if (channel_mode_contains_lr()) { if (b_l_active) { ............................................................... 1 b_l_has_dialog; ............................................................. 1 } if (b_r_active) { ............................................................... 1 b_r_has_dialog; ............................................................. 1 } } if (channel_mode_contains_LsRs()) { b_ls_active; ................................................................... 1 b_rs_active; .................................................................... 1 } if (channel_mode_contains_LrsRrs()) { b_lrs_active; .................................................................. 1 b_rrs_active; ................................................................... 1 } if (channel_mode_contains_LwRw()) { b_lw_active; ................................................................... 1 b_rw_active; .................................................................... 1 } if (channel_mode_contains_VhlVhr()) { b_vhl_active; .................................................................. 1 b_vhr_active; ................................................................... 1 } if (channel_mode_contains_Lfe()) { b_lfe_active; ................................................................... 1 } } if (b_event_probability) { ................................................................ 1 event_probability; ................................................................... 4 } }
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)60
4.2.14.5 drc_frame - DRC frame
Table 70: Syntax of drc_frame()
Syntax No. of bits
drc_frame(b_iframe) { if (b_drc_present) { ...................................................................... 1 if (b_iframe) { drc_config(); } drc_data(); } }
4.2.14.6 drc_config - DRC configuration
Table 71: Syntax of drc_config()
Syntax No. of bits
drc_config() { drc_decoder_nr_modes; ..................................................................... 3 for (m = 0; m <= drc_decoder_nr_modes; m++) { drc_decoder_mode_config(); } drc_eac3_profile; ........................................................................ 3 }
4.2.14.7 drc_decoder_mode_config - DRC decoder mode configuration
Table 72: Syntax of drc_decoder_mode_config()
Syntax No. of bits
drc_decoder_mode_config() { drc_decoder_mode_id; ...................................................................... 3 if (drc_decoder_mode_id > 3) { drc_output_level_from[drc_decoder_mode_id]; .......................................... 5 drc_output_level_to[drc_decoder_mode_id]; ........................................... 5 } if (drc_repeat_profile_flag) { ............................................................ 1 drc_repeat_id[drc_decoder_mode_id]; .................................................. 3 drc_compression_curve_flag[drc_decoder_mode_id] = drc_compression_curve_flag[drc_repeat_id[drc_decoder_mode_id]]; } else { if (!drc_default_profile_flag) { ..................................................... 1 if (drc_compression_curve_flag[drc_decoder_mode_id]) { .......................... 1 drc_compression_curve(); } else { drc_gains_config[drc_decoder_mode_id]; ..................................... 2 } } else { drc_compression_curve_flag[drc_decoder_mode_id] = 1; } } }
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)61
4.2.14.8 drc_compression_curve - Compression curve parameters
Table 73: Syntax of drc_compression_curve()
Syntax No. of bits
drc_compression_curve() { drc_lev_nullband_low; ..................................................................... 4 drc_lev_nullband_high; .................................................................... 4 drc_gain_max_boost; ....................................................................... 4 if (drc_gain_max_boost > 0) { drc_lev_max_boost; ................................................................... 5 drc_nr_boost_sections; ............................................................... 1 if (drc_nr_boost_sections > 0) { drc_gain_section_boost; ......................................................... 4 drc_lev_section_boost; .......................................................... 5 } } drc_gain_max_cut; ......................................................................... 5 if (drc_gain_max_cut > 0) { drc_lev_max_cut; ..................................................................... 6 drc_nr_cut_sections; ................................................................ 1 if (drc_nr_cut_sections > 0) { drc_gain_section_cut; ........................................................... 5 drc_lev_section_cut; ........................................................... 5 } } drc_tc_default_flag; ...................................................................... 1 if (!drc_tc_default_flag) { drc_tc_attack; ....................................................................... 8 drc_tc_release; ...................................................................... 8 drc_tc_attack_fast; .................................................................. 8 drc_tc_release_fast; ................................................................. 8 drc_adaptive_smoothing_flag; ........................................................ 1 if (drc_adaptive_smoothing_flag) { drc_attack_threshold; ........................................................... 5 drc_release_threshold; .......................................................... 5 } } }
4.2.14.9 drc_data -DRC frame-based data
Table 74: Syntax of drc_data()
Syntax No. of bits
drc_data() { curve_present = 0; for (m = 0; m <= 7; m++) { if (drc_compression_curve_flag[m] == 0) { drc_gainset_size = drc_gainset_size_value; ...................................... 6 if (b_more_bits) { .............................................................. 1 drc_gainset_size += variable_bits(2) << 6; } drc_version; ................................................................... 2 used_bits = 0; if (drc_version <= 1) { used_bits = drc_gains(m); } if (drc_version >= 1) { bits_left = drc_gainset_size – 2 – used_bits; drc2_bits; ........................................................ bits_left } } elseif (drc_compression_curve_flag[m] == 1) { curve_present = 1; } } if (curve_present) { drc_reset_flag; ..................................................................... 1 drc_reserved; ....................................................................... 2
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)62
Syntax No. of bits } }
4.2.14.10 drc_gains - DRC gains
Table 75: Syntax of drc_gains()
Syntax No. of bits
drc_gains(mode) { drc_gain[0][0][0] = drc_gain_val; ......................................................... 7 if (drc_gains_config[mode] > 0) { for (ch = 0; ch < nr_drc_channels; ch++) { for (band = 0; band < nr_drc_bands; band++) for (sf = 0; sf < nr_drc_subframes; sf++) { if (sf != 0 || band != 0 || ch != 0) { diff = huff_decode_diff(DRC_HCB, drc_gain_code); ............... 1…x drc_gain[ch][sf][band] = ref_drc_gain + diff; } ref_drc_gain = drc_gain[ch][sf][band]; } ref_drc_gain = drc_gain[ch][0][band]; } ref_drc_gain = drc_gain[ch][0][0]; } } }
NOTE: nr_drc_bands is defined in clause 4.3.13.3.8, nr_drc_channels is defined in clause 4.3.13.7.1, and nr_drc_subframes is defined in clause 4.3.13.7.2.
4.2.14.11 dialog_enhancement - Dialog enhancement metadata
Table 76: Syntax of dialog_enhancement()
Syntax No. of bits
dialog_enhancement(b_iframe) { if (b_de_data_present) { .................................................................. 1 if (b_iframe) { de_config(); } elseif (de_config_flag) { .......................................................... 1 de_config(); } de_data(de_method, de_nr_channels, b_iframe); } }
NOTE: de_nr_channels is defined in table 170.
4.2.14.12 de_config - Dialog enhancement configuration
Table 77: Syntax of de_config()
Syntax No. of bits
de_config() { de_method; ................................................................................ 2 de_max_gain; .............................................................................. 2 de_channel_config; ........................................................................ 3 }
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)63
4.2.14.13 de_data - Dialog enhancement data
Table 78: Syntax of de_data()
Syntax No. of bits
de_data(de_method, de_nr_channels, b_iframe) { if (de_nr_channels > 0) { if ((de_method == 1 || de_method == 3) && de_nr_channels > 1) { if (b_iframe) { de_keep_pos_flag = 0; } else { de_keep_pos_flag; ............................................................. 1 } if (!de_keep_pos_flag) { de_mix_coef1_idx; ............................................................. 5 if (de_nr_channels == 3) { de_mix_coef2_idx; ......................................................... 5 } } } if (b_iframe) { de_keep_data_flag = 0; } else { de_keep_data_flag; ................................................................ 1 } if (!de_keep_data_flag) { if ((de_method==0 || de_method==2) && de_nr_channels==2) { de_ms_proc_flag; .............................................................. 1 } else { de_ms_proc_flag = 0; } for (ch = 0; ch < de_nr_channels - de_ms_proc_flag; ch++) { if (b_iframe && ch == 0) { de_par[0][0] = de_abs_huffman(de_method %2, de_par_code); .......................... 1…x ref_val = de_par[0][0]; de_par_prev[0][0] = de_par[0][0]; for (band = 1; band < de_nr_bands; band++) { de_par[0][band] = ref_val + de_diff_huffman(de_method % 2, de_par_code); .................... 1…x ref_val = de_par[0][band]; de_par_prev[0][band] = de_par[0][band]; } } else { for (band = 0; band < de_nr_bands; band++) { if (b_iframe) { de_par[ch][band] = ref_val + de_diff_huffman(de_method % 2, de_par_code); ............................... 1…x ref_val = de_par[ch][band]; } else { de_par[ch][band] = de_par_prev[ch][band] + de_diff_huffman(de_method % 2, de_par_code); ............................... 1…x } de_par_prev[ch][band] = de_par[ch][band]; } } ref_val = de_par[ch][0]; } if (de_method >= 2) { de_signal_contribution; ....................................................... 5 } } } }
NOTE 1: de_nr_channels is defined in table 170. NOTE 2: de_par_prev contains the parameter indices of the corresponding channel in the previous frame.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)64
4.2.14.14 emdf_payload_config - EMDF payload configuration
Table 79: Syntax of emdf_payload_config()
Syntax No. of bits
emdf_payload_config() { if (b_smpoffst) { ......................................................................... 1 smpoffst = variable_bits(11); } if (b_duration) { ......................................................................... 1 duration = variable_bits(11); } if (b_groupid) { .......................................................................... 1 groupid = variable_bits(2); } if (b_codecdata) { ........................................................................ 1 codecdata; ........................................................................... 8 } if (!b_discard_unknown_payload) { ......................................................... 1 if (b_smpoffst == 0) { if (b_payload_frame_aligned) { .................................................... 1 b_create_duplicate; .......................................................... 1 b_remove_duplicate; .......................................................... 1 } } if (b_smpoffst == 1 || b_payload_frame_aligned == 1) { priority; ........................................................................ 5 proc_allowed; .................................................................... 2 } } }
4.2.14.15 emdf_protection - EMDF protection data
Table 80: Syntax of emdf_protection()
Syntax No. of bits
emdf_protection() { protection_length_primary; ............................................................... 2 protection_length_secondary; ............................................................. 2 protection_bits_primary; .......................................................... 8/32/128 protection_bits_secondary; ...................................................... 0/8/32/128 }
4.3 Description of bitstream elements
4.3.0 Introduction
A number of bitstream elements have values that can be transmitted, but whose meaning has been reserved. If a decoder receives a bitstream that contains reserved values, the decoder can be able to decode and produce audio. In the description of bitstream elements that have reserved codes, there is an indication of what the decoder can do if the reserved code present. In some cases, the decoder cannot decode audio. In other cases, the decoder can still decode audio by using a default value for a parameter which was indicated by a reserved code.
In many cases, the following text provides interpretations for the bits transmitted in the bitstream, by formulas or tables. When a bitstream element is referred to later in the document for an element where such an interpretation is given, the reference is understood to refer to the interpreted value, not to the bits.
EXAMPLE 1: When wait_frames is referred to in clause 5 and later, wait_frames = 4 means that a bit pattern of 011 has been transmitted if the frame_rate_index value was 10.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)65
If bitstreams elements with the same name and meaning are present in multiple syntactical elements, the description of this element is present only once.
EXAMPLE 2: max_sfb[0] is present with the same meaning in sf_info_lfe() and asf_psy_info(). The element is documented only as part of the description of the asf_psy_info() element.
4.3.1 raw_ac4_frame - raw AC-4 frame
4.3.1.1 fill_area - fill area
This element is used to pad the raw_ac4_frame to a certain transmission frame size.
Padding can occur before the first ac4_substream_data or after the last ac4_substream_data. This split of the padding bits can be beneficial for editing purposes. The start of the first ac4_substream_data is given by the payload base offset.
4.3.1.2 fill_bits - byte alignment bits
This element can be used by an AC-4 encoder to achieve an average bit rate (ABR) stream. Decoders should ignore this element.
4.3.1.3 byte_align - byte alignment bits
This element is used for the byte alignment of the raw_ac4_frame() element. Byte alignment is defined relative to the start of the enclosing syntactic element.
4.3.2 variable_bits - Variable bits
4.3.2.0 Encoding
The variable_bits() method is an extension mechanism for variable bit-length elements to use more bits for the element than the minimum element length. This method enables efficient coding of small field values with extensibility to be able to express arbitrarily large field values. Field values are split into one or more groups of n_bits bits, with each group followed by the 1-bit b_read_more field. At a minimum, coding of the value requires n_bits + 1 bit to be transmitted. All fields coded using variable_bits() shall be interpreted as unsigned integers.
If the value to be encoded is between 0 and 2n_bits - 1, then only one group of n_bits bits is required. The field value is in this case equal to the value to be encoded, and the b_read_more field shall be set to 0. If the value to be encoded is larger than 2n_bits - 1, the b_read_more field is set to 1, and a second group of n_bits follows in the bitstream, followed by another Boolean. If the value to be encoded falls within the range [2n_bits, 2n_bits + 22×n_bits - 1], then the second Boolean is set to false. Otherwise, it is set to true and a third group of n_bits follows. This process continues until a word of sufficient length is transmitted.
4.3.2.1 read - read bits
This element is used to determine the value of variable_bits as specified by the syntax box in clause 4.2.2.
The size of this bit field is specified by the parameter n_bits.
4.3.2.2 b_read_more - read more flag
If true, this Boolean indicates indicates the presence of additional read bits.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)66
4.3.3 AC-4 frame information
4.3.3.1 Purpose
This element contains the configuration for an AC-4 decoder. An AC-4 decoder shall only be configured by the ac4_toc() (as opposed to information available on system level).
4.3.3.2 ac4_toc - AC-4 table of contents
4.3.3.2.1 bitstream_version - bitstream version
This element indicates the bitstream version of the AC-4 codec frame. It is a 2-bit element that can be extended by variable_bits().
Only bitstreams with a bitstream version of 0 and 1 are decodable using this version of the AC-4 specification. Bitstreams with a bitstream version of 2 or higher are not decodable according to the present document. Encoders conforming to the present document shall use a value of 0 for bitstream_version.
4.3.3.2.2 sequence_counter - sequence counter
This element contains the sequence counter value for the current AC-4 frame. Using the sequence counter, a decoder can detect uncontrolled changes in the stream source, such as might occur in splicing operation.
Normal operation A decoder may continue its usual processing if one the following conditions is met:
• Counter increase: sequence_counter = sequence_counter_prev + 1
• Counter wrap around: sequence_counter = 1 AND sequence_counter_prev = 1 020
• Splice in previous frame: sequence_counter ≠ 0 AND sequence_counter_prev = 0
Here, sequence_counter_prev indicates the sequence_counter value from the previous AC-4 frame.
Source change detected If none of the conditions above apply, the decoder detects a source change in the current frame. Although the present document does not define exact decoder behaviour at source changes, a decoding system should provide continuity of audio experience in the period between recognition of a source change and the next I-Frame.
A splicing device should overwrite the sequence_counter value of the first frame after a splice with 0.
An encoder conforming to the present document shall not write sequence_counter values exceeding 1 020.
4.3.3.2.3 b_wait_frames
If true, this Boolean indicates that wait_frames information shall follow in the bitstream. If false, wait_frames is undefined.
4.3.3.2.4 wait_frames
For ABR streams, this element indicates when decoding of this frame can begin. After having received the frame carrying wait_frames, a decoding system should wait for wait_frames frames before the decoded frame is placed into the output buffer, unless other timing information exists. Refer to table 81 for the interpretation of the bits.
NOTE 1: This requirement makes sure that if the bitstream is received over a constant-rate channel, the decoder input buffer does not run dry before the next output frame is due.
For constant bit rate (CBR) streams where every frame carries the same number of bits, the decoder need not wait at all. For variable bit rate (VBR) streams, the delay is unknown.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)67
NOTE 2: VBR streams are not well suited for transmission over a constant rate channel.
When the delay is unknown, the decoder should assume audio continuity, unless a source change was detected. In that case, the maximum delay of 5 (10) frames should be assumed.
Table 81: Decoding delay
wait_frames Number of frames to wait before output frame_rate_index = [0...9, 13] frame_rate_index = [10, 11, 12]
0 stream is CBR (=0) 1 0 0 2 1 2 3 2 4 4 3 6 5 4 8 6 5 10 7 Stream is VBR; no statement on wait frames can be made.
4.3.3.2.5 fs_index - Sampling frequency index
This element indicates the base sampling frequency, identified by the variable base_samp_freq, as shown in table 82.
Table 82: Base sampling frequency
fs_index base_samp_freq (kHz) 0 44,1 1 48
4.3.3.2.6 frame_rate_index - frame rate
This element indicates the rate of frames, frame_rate, in units of 1/s, the value of frame_length indicating the length of the frame in samples (dependent on the external sample rate) and the decoder resampling ratio as shown in table 83 and table 84. The base codec frame length value, frame_len_base, is defined as the frame length at an external sample rate of 48 kHz.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)68
Table 83: frame_rate_index for 48 kHz, 96 kHz, and 192 kHz
frame_rate_index frame_rate (fps)
Internal frame length (samples) frame_length Decoder resampling ratio for external 48 kHz =
frame_len_base for external
96 kHz for external
192 kHz 0 23,976 1 920 3 840 7 680 1001 1000⁄
× 25 24⁄ 1 24 1 920 3 840 7 680 25 24⁄ 2 25 2 048 4 096 8 192 15 16⁄ 3 29,97 1 536 3 072 6 144 1001 1000⁄
× 25 24⁄ 4 30 1 536 3 072 6 144 25 24⁄ 5 47,95 960 1 920 3 840 1001 1000⁄
× 25 24⁄ 6 48 960 1 920 3 840 25 24⁄ 7 50 1 024 2 048 4 096 15 16⁄ 8 59,94 768 1 536 3 072 1001 1000⁄
× 25 24⁄ 9 60 768 1 536 3 072 25 24⁄ 10 100 512 1 024 2 048 15 16⁄ 11 119,88 384 768 1 536 1001 1000⁄
× 25 24⁄ 12 120 384 768 1 536 25 24⁄ 13 (23,44) 2 048 4 096 8 192 1 14: reserved - - - - - 15: reserved - - - - -
Table 84: frame_rate_index for 44,1 kHz
frame_rate_index frame_rate (fps)
Internal frame length (samples) frame_length (for external 44,1 kHz = frame_len_base)
Decoder resampling ratio
0…12: reserved - - - 13 11025 512⁄ 2 048 1 14: reserved - - - 15: reserved - - -
4.3.3.2.7 b_iframe_global - global I-frame flag
If true, this Boolean indicates that all substreams in all presentations have b_iframe set to true.
If frame_rate_factor!=1, this is fulfilled if the first b_iframe of a series of 2 or 4 substreams is true.
4.3.3.2.8 b_single_presentation - single presentation flag
If true, this Boolean indicates the presence of a single presentation. If false, b_more_presentations and additional variable_bits are used to derive the number of presentations contained in the AC-4 frame.
4.3.3.2.9 b_more_presentations - more presentations flag
If true, this Boolean indicates that the number of presentations, n_presentations, is derived from additional variable_bits. If false, no presentation is contained in the AC-4 frame.
4.3.3.2.10 b_payload_base - payload base flag
If true, this Boolean indicates that payload base offset information shall follow in the bitstream. If false, the payload base offset is 0 and the first ac4_substream_data shall immediately follow the ac4_toc element.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)69
4.3.3.2.11 payload_base_minus1 - payload base offset minus 1
This element indicates the start of the ac4_substream_data for substream 0 relative to the end of the byte-aligned ac4_toc element in bytes minus 1. It is a 5-bit element that is extended by variable_bits() for a value of 0x1f.
4.3.3.2.12 byte_align - byte alignment bits
This element is used for the byte alignment of the ac4_toc() element.
4.3.3.3 ac4_presentation_info - AC-4 presentation information
4.3.3.3.1 b_single_substream - single substream flag
If true, this Boolean indicates that the presentation contains a single substream.
4.3.3.3.2 b_belongs_to_presentation_id - Presentation identifier assignment flag
If true, this Boolean indicates that the audio presentation belongs to a presentation identifier. The presentation identifier presentation_id is derived from additional variable_bits.
4.3.3.3.3 b_hsf_ext - high sampling frequency extension flag
If true, this Boolean indicates that additional spectral data is available that can be used for decoding a presentation into 96 kHz or 192 kHz.
4.3.3.3.4 presentation_config - presentation configuration -
This element indicates the presentation configuration as shown in table 85. It is 3-bit element that is extendable by variable_bits().
Table 85: presentation_config
presentation_config Presentation configuration 0 music and effects (M+E) + dialogue 1 Main + dialogue enhancement 2 Main + associate 3 Music and effects (M+E) + dialogue enhancement + associate 4 Main + dialogue enhancement + associate 5 HSF extended ≥ 6 Reserved
4.3.3.3.5 b_pre_virtualized - pre-virtualized flag
If true, this Boolean indicates that the audio content in the current presentation was pre-rendered by a headphone virtualizer. In this case, the receiving device/service may choose to turn off device-side content postprocessing.
4.3.3.3.6 b_add_emdf_substreams - additional EMDF substreams flag
If true, this Boolean indicates the presence of additional EMDF substreams.
4.3.3.3.7 n_add_emdf_substreams - number of additional EMDF substreams
This element indicates the number of additional EMDF substreams. It is a 2-bit element that is extendable by variable_bits().
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)70
4.3.3.3.8 mdcompat - compatibility indication
This field indicates the decoder compatibility as shown in table 86. The mdcompat element indicates decoder systems that are compatible with a presentation. A system with compatibility level n shall decode all presentations with mdcompat ≤ n. A system with compatibility level n should not decode (i.e. select) presentations with mdcompat > n.
Table 86: mdcompat
mdcompat Maximum number of channels (including Low-Frequency Effects (LFE))
Maximum channel configuration of
main or M+E
dialogue associate
0 2 2 n/a n/a 1 6 5.1 n/a n/a 2 9 5.1 3 2 3 11 5.1 3 2 4 13 7.1 3 2 5-6 Reserved 7 Unrestricted
4.3.3.4 presentation_version - presentation version information
4.3.3.4.1 b_tmp - temporary flag
One or more sequential instances of this element signal the version of the presentation (i.e. it might be present multiple times).
An encoder conforming to the present document shall write a presentation version value of 0. A decoder implemented in accordance with the present document shall not decode a presentation if the version of the presentation is not 0.
4.3.3.5 frame_rate_multiply_info - frame rate multiplier information
4.3.3.5.1 b_multiplier - multiplier flag
If true, this Boolean indicates a frame_rate_factor value different to 1. If FALSE or not present, the frame_rate_factor value is 1. See clause 4.3.3.5.3.
4.3.3.5.2 multiplier_bit - multiplier bit
If set, this bit indicates a frame_rate_factor value of 4 and if not set, a frame_rate_factor value of 2. See clause 4.3.3.5.3.
4.3.3.5.3 frame_rate_factor - frame rate factor
The value of this element, as specified in table 87, determines whether an ac4_substream_info() element refers to 1, 2, or 4 substreams. Each of those substreams shall be decoded consecutively.
Table 87: frame_rate_factor
frame_rate_index b_multiplier multiplier_bit frame_rate_factor 2, 3, 4 0 X 1
1 0 2 1 1 4
0, 1, 7, 8, 9 0 X 1 1 X 2
5, 6, 10, 11, 12, 13 X X 1
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)71
4.3.3.6 emdf_info - EMDF information
4.3.3.6.1 emdf_version - EMDF syntax version
This element indicates the syntax version that the EMDF substream conforms with. It is a 2-bit element that can be extended by variable_bits(). For substreams that conform to the syntax defined in this version of the AC-4 specification, the emdf_version field shall be set to 0.
4.3.3.6.2 key_id - authentication ID
The value this element identifies the hashing algorithm used to calculate the value of the protection_bits_primary and protection_bits_secondary fields of the emdf_protection element. It is a 3-bit element that can be extended by variable_bits().
The decoder shall ignore protection_bits_primary and protection_bits_secondary if the key_id is equal to 0x06; else it may ignore protection_bits_primary and protection_bits_secondary.
NOTE: All other values of key_id are implementation dependent and are not defined in the present document.
4.3.3.6.3 b_emdf_payloads_substream_info - EMDF payloads substream information flag
If true, This Boolean indicates the presence of an emdf_payloads_substream_info() element.
4.3.3.7 ac4_substream_info - AC-4 substream information
4.3.3.7.1 channel_mode - channel mode
This element indicates the channel mode and the ch_mode variable as shown in table 88. The element has a variable length of 1, 2, 4, or 7 bits that can be extended by variable_bits().
Table 88: channel_mode
Value of channel_mode Channel mode ch_mode 0 Mono 0 10 Stereo 1 1100 3.0 2 1101 5.0 3 1110 5.1 4 1111000 7.0: 3/4/0 (L, C, R, Ls, Rs, Lrs, Rrs) 5 1111001 7.1: 3/4/0.1 (L, C, R, Ls, Rs, Lrs, Rrs, LFE) 6 1111010 7.0: 5/2/0 (L, C, R, Lw, Rw, Ls, Rs) 7 1111011 7.1: 5/2/0.1 (L, C, R, Lw, Rw, Ls, Rs, LFE) 8 1111100 7.0: 3/2/2 (L, C, R, Ls, Rs, Vhl, Vhr) 9 1111101 7.1: 3/2/2.1 (L, C, R, Ls, Rs, Vhl, Vhr, LFE) 10 1111110 Reserved 11 ≥ 1111111 Reserved 12+
The 3.0 channel mode shall only be used in the following context:
• for coding of the enhancement signal for the dialogue enhancement (DE) feature;
• for coding of the dialogue in a music and effects (M+E) + dialogue presentation.
4.3.3.7.2 b_sf_multiplier - sampling frequency multiplier flag
If true, this Boolean indicates that the sampling frequency of the AC-4 substream is a multiple of the base sampling frequency. If false, the sampling frequency of the AC-4 substream is identical to the base sampling frequency.
NOTE: b_sf_multiplier is only available if the base sampling frequency is 48 kHz.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)72
4.3.3.7.3 sf_multiplier - sampling frequency multiplier bit
This element indicates the sampling frequency multiplier. Since this bit is only available if the base sampling frequency is 48 kHz, the sampling frequency of the AC-4 substream is given by table 89.
Table 89: AC-4 substream sampling frequency for a base sampling frequency of 48 kHz
b_sf_multiplier sf_multiplier Sampling frequency 0 - 48 kHz 1 0 96 kHz
1 192 kHz
4.3.3.7.4 b_bitrate_info - bit-rate presence flag
If true, the Boolean indicates that a bit rate indicator is specified.
4.3.3.7.5 bitrate_indicator - bit-rate indicator
This element indicates the upper average bit-rate limit per channel in the substream and the variable brate_ind as shown in table 90.
Table 90: bitrate_indicator
Value of bitrate_indicator Bitrate (kbit/s) brate_ind 0b000 16 0 0b010 20 1 0b100 24 2 0b110 28 3 0b00100 32 4 0b00101 40 5 0b00110 48 6 0b00111 56 7 0b01100 64 8 0b01101 80 9 0b01110 96 10 0b01111 112 11 0b1X1XX (8 further values) Unlimited 12…19
4.3.3.7.6 add_ch_base - additional channels coupling base
This element indicates whether the A-CPL coding of the additional channels (Lw/Rw or Vhl/Vhr depending on channel_mode) is based on the L/R pair (add_ch_base = 0) or on the Ls/Rs pair (add_ch_base = 1). See clause 5.7.7.6.3 for details.
4.3.3.7.7 b_content_type - content type presence flag
If true, this Boolean indicates that a content type is specified.
4.3.3.7.8 b_iframe - I-frame flag
This Boolean is set for each of the frame_rate_factor substreams. If true, it indicates that the contained frame is an I-frame.
4.3.3.7.9 substream_index - substream index
This element indicates the substream index that the ac4_substream_info() element refers to. It is a 2-bit element that can be extended by variable_bits(). The substream index is used as index in the substream_index_table() element to get the offset to ac4_substream(). If frame_rate_factor is not 1, substream_index refers to the first of frame_rate_factor substreams in substream_index_table().
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)73
NOTE: Each of the frame_rate_factor substreams has an entry in substream_index_table(). The consecutive entries of frame_rate_factor are stored in substream_index_table() starting at substream_index.
4.3.3.8 content_type - content type
4.3.3.8.1 content_classifier - content classifier
This element classifies the content of an audio program component as shown in table 91.
Table 91: content_classifier
Value of content_classifier Content classification (audio program component type) 000 Main audio service: complete main 001 Main audio service: music and effects 010 Associated service: visually impaired 011 Any associated service: hearing impaired 100 Associated service: dialogue 101 Any associated service: commentary 110 Associated service: emergency 111 Associated service: voice over
4.3.3.8.2 b_language_indicator - programme language indicator flag
If true, this Boolean indicates that programme language indication data is available.
4.3.3.8.3 b_serialized_language_tag - serialized language tag flag
The Boolean b_serialized_language_tag field indicates how the language tag is delivered:
• If true, the language tag is delivered using multiple payloads in a sequence of AC-4 frames.
• If false, the complete language tag is delivered in the current payload.
The value of b_serialized_language_tag shall be set to 1 in all programme language payloads that follow the programme language payload in which both the value of the b_serialized_language_tag field and the value of the b_start_tag field is set to true, up to and including the payload that contains the final byte of the language tag.
4.3.3.8.4 b_start_tag - language tag start flag
The Boolean b_start_tag field indicates the start of a multi-frame sequence of language tag data, as follows:
• If true, the payload contains the first chunk of the language tag.
Decoders shall start decoding the language tag beginning with the data in this payload.
• If false, the payload does not contain the start of the multi-frame sequence.
4.3.3.8.5 language_tag_chunk - language tag chunk
This element is used to identify the program language as defined in IETF BCP 47 [4].
The element shall contain a two-byte section of a language tag that conforms to the syntax and semantics defined in IETF BCP 47 [4]. The most significant byte of the language tag shall be stored in the most significant byte of the language_tag_chunk in a programme language payload that has a b_serialized_language_tag value of true. Subsequent bytes of the language tag shall be stored in the second byte of the current language_tag_chunk and in the language_tag_chunk of each of the subsequent programme language payloads required to deliver the complete language tag. If the length of the language tag does not correspond to an even number of bytes, the value of the extra byte that follows the end of the language tag shall be set to 0x0.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)74
4.3.3.8.6 n_language_tag_bytes - number of language tag bytes
This element indicates the total length, in bytes, of the language tag. Values of 0 and 1 and values of 43 to 63 are reserved.
4.3.3.8.7 language_tag_bytes - language tag bytes
This element identifies the language of the program.
The sequence of language_tag_bytes shall contain a language tag that conforms to the syntax and semantics defined in IETF BCP 47 [4]. The minimum sequence length shall be two bytes, supporting a two-character language tag as specified in ISO 639-1 [5]. The maximum supported length of the language tag shall be 336 bits or 42 bytes.
4.3.3.9 presentation_config_ext_info - presentation configuration extended information
4.3.3.9.1 n_skip_bytes - number of bytes to skip
This element specifies the number of bytes to skip.
4.3.3.9.2 b_more_skip_bytes - more bytes to skip flag
If true, this Boolean indicates that the number of bytes to skip, n_skip_bytes, is extended by the use of variable_bits().
4.3.3.9.3 reserved - Reserved
This element is reserved for future use and the content shall be skipped by an AC-4 decoder conforming to the present document.
4.3.3.10 ac4_hsf_ext_substream_info - AC-4 HSF extension substream information
4.3.3.10.1 substream_index - substream index
This element indicates the substream index that the ac4_hsf_ext_substream_info element refers to. It is a 2-bit element that can be extended by variable_bits(). The substream index is used as index in substream_index_table() to get the offset to ac4_substream() for the high sampling frequency extension substream.
4.3.3.11 emdf_payloads_substream_info - EMDF payloads substream information
4.3.3.11.1 substream_index - substream index
This element indicates the substream index that the emdf_payloads_substream_info element refers to. It is a 2-bit element that can be extended by variable_bits(). The substream index is used as index in substream_index_table() to get the offset to emdf_payloads_substream().
4.3.3.12 substream_index_table - substream index table
4.3.3.12.1 n_substreams - number of substreams
This element indicates the number of substreams available in substream_index_table. It is a 2-bit element that can be extended by variable_bits().
4.3.3.12.2 b_size_present - size present flag
If true, this Boolean indicates the presence of the substream size.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)75
4.3.3.12.3 b_more_bits - more bits flag
If true, this Boolean indicates that additional variable_bits() are used to determine the substream size.
4.3.3.12.4 substream_size - substream size
This element indicates the substream size in bytes for the substream with index s. It is a 10-bit element that can be extended by variable_bits().
To get the offset for the nth substream relative to the end of ac4_toc, the substream sizes of the substreams with a substream index less than n need to be accumulated as shown in the following pseudocode.
Pseudocode 1
// get the substream offset for substream with substream index n substream_n_offset = payload_base; for (s = 0; s < n; s++) { substream_n_offset += substream_size[s]; }
4.3.4 ac4_substream - AC-4 substream
4.3.4.1 audio_size_value - audio size value
This element indicates the audio_size in bytes. It is a 15-bit element that can be extended by variable_bits(). The audio_size is the size of the audio_data element including any following fill_bits and byte_align bits but not including the size of the metadata element and the final byte_align bits.
NOTE: This enables decoders to directly access metadata without the need to parse audio data.
4.3.4.2 b_more_bits - more bits flag
If true, this Boolean indicates the presence of additional variable_bits to be used for the audio_size determination.
4.3.4.3 byte_align - byte alignment bits
This element is used for byte alignment within the ac4_substream() element. An encoder shall use this field to pad the length of ac4_substream to an integer number of bytes.
4.3.5 Channel elements
4.3.5.0 Introduction
This clause describes the bitstream elements that can be found in single_channel_element, mono_data, channel_pair_element, stereo_data, 5_X_channel_element, and 7_X_channel_element.
4.3.5.1 mono_codec_mode - mono codec mode
The value of this element indicates the mono codec mode as shown in table 92.
Table 92: Mono codec mode
mono_codec_mode Mono codec mode 0 = SIMPLE Simple 1 = ASPX advanced spectral extension
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)76
4.3.5.2 spec_frontend - spectral frontend selection
The value of this element indicates the used spectral frontend as shown in table 93.
Table 93: Spectral frontend selection
spec_frontend Spectral frontend 0 = ASF audio spectral front end 1 = SSF Speech Spectral Frontend
The bitstream elements spec_frontend_m, spec_frontend_s, spec_frontend_l, and spec_frontend_r are also spectral frontend selection flags with the same meaning. The extension indicates the channel type: m = mid, s = side, l = left and r = right.
4.3.5.3 stereo_codec_mode - stereo codec mode
The value of this element indicates the stereo codec mode as shown in table 94.
Table 94: stereo_codec_mode
stereo_codec_mode Stereo codec mode 0 = SIMPLE Simple 1 = ASPX A-SPX 2 = ASPX_ACPL_1 A-SPX, A-CPL mode 1 3 = ASPX_ACPL_2 A-SPX, A-CPL mode 2
4.3.5.4 3_0_codec_mode - 3.0 codec mode
The value of this element indicates the 3.0 codec mode as shown in table 95.
Table 95: 3_0_codec_mode
3_0_codec_mode 3.0 codec mode 0 = SIMPLE Simple 1 = ASPX advanced spectral extension (A-SPX)
4.3.5.5 3_0_coding_config - 3.0 coding configuration
This element indicates the coding configuration for the 3.0 channel mode.
4.3.5.6 5_X_codec_mode - 5.X codec mode
The value of this element indicates the 5.X codec mode as shown in table 96.
Table 96: 5_X_codec_mode
5_X_codec_mode 5.X codec mode 0 = SIMPLE Simple 1 = ASPX A-SPX 2 = ASPX_ACPL_1 A-SPX, A-CPL mode 1 3 = ASPX_ACPL_2 A-SPX, A-CPL mode 2 4 = ASPX_ACPL_3 A-SPX, A-CPL mode 3 5…7 Reserved
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)77
4.3.5.7 7_X_codec_mode - 7.X codec mode
The value of this element indicates the 7.X codec mode as shown in table 97.
Table 97: 7_X_codec_mode
7_X_codec_mode 7.X codec mode 0 = SIMPLE Simple 1 = ASPX A-SPX 2 = ASPX_ACPL_1 A-SPX, A-CPL mode 1 3 = ASPX_ACPL_2 A-SPX, A-CPL mode 2
4.3.5.8 coding_config - coding configuration
This element indicates the coding configuration for the 5.X or the 7.X channel mode.
4.3.5.9 2ch_mode - channel coupling mode
The value of this element indicates the way channel pairs are coupled into the output channels. See clause 5.3.4.3 and clause 5.3.4.4.0 for details.
4.3.5.10 b_enable_mdct_stereo_proc - enable MDCT stereo processing flag
If true, this Boolean indicates the presence of MDCT domain stereo processing data stored in a chparam_info element. The sf_info element, which is also enabled by this Boolean, holds audio spectral front end (ASF) information that is valid for the decoding of both sf_data elements that follow.
4.3.5.11 chel_matsel - matrix selection code
This element indicates the matrix for multichannel coupling.
4.3.5.12 b_use_sap_add_ch - use SAP for additional channels flag
If true, this Boolean indicates that the SAP tool shall be used for the additional channels and that SAP data stored in two chparam_info() elements shall follow in the bitstream.
4.3.5.13 max_sfb_master - max_sfb indication for related channels
This element indicates the max_sfb value to be used when decoding the two sf_data() elements, which follow the max_sfb_master element and the two chparam_info() elements. The max_sfb value for various block sizes is shown in table B.8 to table B.19. The table to use depends on the largest signalled transform length which is also used for the n_side_bits determination as indicated in the notes in clause 4.2.6.6 and clause 4.2.6.14. If the transform length for a block is equal to the largest signalled transform length, the max_sfb_master value maps directly to the max_sfb value. Otherwise, the max_sfb value is given by the n_sfb_side value for the transform length of the block and the max_sfb_master value.
4.3.6 audio spectral front end
4.3.6.1 asf_transform_info - ASF transform information
4.3.6.1.1 b_long_frame - long frame flag
If true, this Boolean indicates that the audio frame is a long frame.
A long frame contains just one (full) block that needs to be transformed. The transform length for a long frame depends on the sampling frequency and the frame_len_base value and is given by table 98.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)78
Table 98: Transform length for long frames
frame_len_base Sampling frequency 44,1 kHz, 48 kHz 96 kHz 192 kHz
2 048 2 048 4 096 8 192 1 920 1 920 3 840 7 680 1 536 1 536 3 072 6 144 1 024 1 024 2 048 4 096 960 960 1 920 3 840 768 768 1 536 3 072 512 512 1 024 2 048 384 384 768 1 536
4.3.6.1.2 transf_length[i] - transform length index i
The value of this element is used to derive the transform length for partial blocks if the frame_len_base value is greater or equal to 1 536.
An audio frame contains several partial blocks if the audio frame is not a long frame. All partial blocks that make up the first half of a corresponding full block use index 0 and the others use index 1. Table 99 through table 101 show the transform length values for partial blocks depending on the sampling frequency and the frame_length value.
NOTE: The transf_length[i] values are only available if b_long_frame is false. If b_long_frame is true, the transform length for the audio block equals to the frame_length value.
Table 99: Transform length for non-long frames, frame_len_base ≥ 1 536 and 44,1 kHz or 48 kHz
frame_length transf_length[i] 0 1 2 3
2 048 128 256 512 1 024 1 920 120 240 480 960 1 536 96 192 384 768
Table 100: Transform length for non-long frames, frame_len_base ≥ 1 536 and 96 kHz
frame_length transf_length[i] 0 1 2 3
4 096 256 512 1 024 2 048 3 840 240 480 960 1 920 3 072 192 384 768 1 536
Table 101: Transform length for non-long frames, frame_len_base ≥ 1 536 and 192 kHz
frame_length transf_length[i] 0 1 2 3
8 192 512 1 024 2 048 4 096 7 680 480 960 1 920 3 840 6 144 384 768 1 536 3 072
4.3.6.1.3 transf_length - transform length
The value of this element is used to derive the transform length for the audio block or the partial blocks if the frame_len_base value is less than 1 536. Table 102 through table 104 show the transform length values depending on the sampling frequency and the frame_length value.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)79
Table 102: Transform length for frame_len_base < 1 536 and 44,1 kHz or 48 kHz
frame_length transf_length - 0 1 2 3 1 024 128 256 512 1 024 960 120 240 480 960 768 96 192 384 768 512 128 256 512 × 384 96 192 384 ×
Table 103: Transform length for frame_len_base < 1 536 and 96 kHz
frame_length transf_length - 0 1 2 3 2 048 256 512 1 024 2 048 1 920 240 480 960 1 920 1 536 192 384 768 1 536 1 024 256 512 1 024 × 768 192 384 768 ×
Table 104: Transform length for frame_len_base < 1 536 and 192 kHz
frame_length transf_length - 0 1 2 3 4 096 512 1 024 2 048 4 096 3 840 480 960 1 920 3 840 3 072 384 768 1 536 3 072 2 048 512 1 024 × × 1 536 384 768 × ×
4.3.6.1.4 get_transf_length(g) – get transform length for group g
This helper function returns the transf_length value for the window group g.
Pseudocode 2
get_transf_length(g) { if (frame_len_base >= 1536) { if (b_long_frame == 0) { num_windows_0 = (1 << (3-transf_length[0])); if (g < window_to_group[num_windows_0]) { return transf_length[0]; } else { return transf_length[1]; } } else { return 4; // long frame, the transform length equals to frame_length } } else { return transf_length; } }
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)80
4.3.6.2 asf_psy_info - ASF psy information
4.3.6.2.1 n_msfb_bits - number of maxsfb bits
The number of bits used for the max_sfb[i] and max_sfb_side[i] elements depends on the transform length and is determined by the value of n_msfb_bits in table 105 through table 107. The number of bits for max_sfb[0] in the sf_info_lfe() element is determined by the value of n_msfbl_bits in table 105 through table 107.
In addition to specifying the n_msfb_bits and n_msfbl_bits variables, table 105 also specifies the n_side_bits variable, which indicates a codeword length as follows:
• if b_side_limited is true, n_side_bits indicates the number of bits used for the max_sfb_side[i] element; or
• if max_sfb_master is present, n_side_bits indicates the number of bits for the max_sfb_master element.
If a high sampling frequency extension is not present in the presentation, which is signalled by b_hsf_ext = 0, the transform length indicated in table 105 shall be used. If b_hsf_ext = 1, the transform length for the high sampling frequency shall be used as indicated in either table 106 or table 107.
Table 105: n_msfb_bits and n_side_bits for 44,1 kHz or 48 kHz
Transform length n_msfb_bits n_side_bits n_msfbl_bits 2 048 6 5 3 1 920 6 5 3 1 536 6 5 3 1 024 6 5 2 960 6 5 2 768 6 5 2 512 6 5 2 480 6 5 n/a 384 6 4 2 256 5 4 n/a 240 5 4 n/a 192 5 3 n/a 128 4 3 n/a 120 4 3 n/a 96 4 3 n/a
Table 106: n_msfb_bits for 96 kHz
Transform length n_msfb_bits n_msfbl_bits 4 096 6 3 3 840 6 3 3 072 6 3 2 048 6 2 1 920 6 2 1 536 6 2 1 024 6 2 960 6 n/a 768 6 2 512 5 n/a 480 5 n/a 384 5 n/a 256 4 n/a 240 4 n/a 192 4 n/a
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)81
Table 107: n_msfb_bits for 192 kHz
Transform length n_msfb_bits n_msfbl_bits 8 192 6 3 7 680 6 3 6 144 6 3 4 096 6 2 3 840 6 2 3 072 6 2 2 048 6 2 1 920 6 n/a 1 536 6 2 1 024 5 n/a 960 5 n/a 768 5 n/a 512 4 n/a 480 4 n/a 384 4 n/a
4.3.6.2.2 max_sfb[i] - number of transmitted scale factor bands for index i
This element is used for a full block or all partial blocks related to transf_length[0] and max_sfb[1] is used for all partial blocks related to transf_length[1].
Depending on the value of the transform length, this is a 4-bit, 5-bit or 6-bit integer value indicating the number of transmitted scale factor bands in asf_psy_info. The number of bits used for max_sfb[i] is given in clause 4.3.6.2.1. The value is less or equal to num_sfb.
4.3.6.2.3 max_sfb_side[i] - number of transmitted scale factor bands for side channel and index i
This element indicates the number of transmitted scale factor bands for the second (i.e. side) channel. This value is typically only transmitted when the second channel has a different number of scale factor bands than the first channel. If no max_sfb_side[i] is transmitted, max_sfb[i] is also used for the second channel. This variable lenght element has a size of 3, 4 or 5 bits.
4.3.6.2.4 n_grp_bits - number of grouping bits
The number of grouping bits is 0 if b_long_frame is true and the frame_len_base value is greater or equal to 1 536. If b_long_frame is false and the frame_len_base value is greater or equal to 1 536, the number of grouping bits n_grp_bits depends on transf_length[0], transf_length[1], and the frame_len_base value, and is given by table 108. If the frame_len_base value is less than 1 536, the number of grouping bits, n_grp_bits, depends on transf_length and the frame_len_base value and is given by table 109.
Table 108: n_grp_bits for frame_len_base ≥ 1 536 and b_long_frame=0
frame_len_base transf_length[0] transf_length[1] n_grp_bits 2 048, 1 920, 1 536 0 0 15
0 1 10 0 2 8 0 3 7 1 0 10 1 1 7 1 2 4 1 3 3 2 0 8 2 1 4 2 2 3 2 3 1 3 0 7 3 1 3
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)82
frame_len_base transf_length[0] transf_length[1] n_grp_bits 3 2 1 3 3 1
Table 109: n_grp_bits for frame_len_base < 1 536
frame_len_base transf_length n_grp_bits 1 024, 960, 768 0 7 - 1 3 - 2 1 - 3 0 512, 384 0 3 - 1 1 - 2 0
4.3.6.2.5 scale_factor_grouping_bit - scale factor grouping bit
This element, which is present n_grp_bits times, is used to indicate the scale factor grouping. The conversion of the scale_factor_grouping[i] array into asf_psy_info helper elements is specified in clause 4.3.6.2.6.
4.3.6.2.6 asf_psy_info helper elements
The following helper elements are derived from the asf_psy_info bitstream elements, especially from the scale_factor_grouping[i] array:
num_windows Number of (transform) windows within the current audio frame. Possible values are: 1, 2, 3, 4, 5, 6, 8, 9, 10, 12, and 16.
num_window_groups
Number of window groups. window_to_group[w]
Array of size num_windows that maps a window number w into a group number g. num_win_in_group[g]
Vector indicating the number of windows in group g. sect_sfb_offset[g][sfb]
Array indicating the section scale factor band offset for the scale factor band sfb within group g.
NOTE: If transf_length[1] = transf_length[0], num_windows = 2, 4, 8 or 16 and if transf_length[1] ≠ transf_length[0], num_windows = 3, 5, 6, 9, 10, or 12. If b_long_frame is true, num_windows = 1.
Pseudocode 3
// derive num_windows, num_window_groups and window_to_group[w] num_windows = 1; num_window_groups = 1; window_to_group[0] = 0; if (b_long_frame == 0) { num_windows = n_grp_bits + 1; if (b_different_framing) { /* nr of windows in first half of frame */ num_windows_0 = (1 << (3-transf_length[0])); for (i = n_grp_bits; i >= num_windows_0; i--) { /* shift grouping bits of 2nd half by 1 */ scale_factor_grouping[i] = scale_factor_grouping[i-1]; } /* no grouping for unequal transform lengths */ scale_factor_grouping[num_windows_0-1] = 0; num_windows++; }
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)83
for (i = 0; i < num_windows - 1; i++) { if (scale_factor_grouping[i] == 0) { num_window_groups += 1; } window_to_group[i + 1] = num_window_groups - 1; } }
Pseudocode 4
// derive num_win_in_group[g] and sect_sfb_offset[g][sfb] group_offset = 0; for (g = 0; g < num_window_groups; g++) { num_win_in_group[g] = 0; for (win = 0; win < num_windows; win++) { if (window_to_group[win] == g) { num_win_in_group[g] += 1; } } max_sfb = get_max_sfb(g); for (sfb = 0; sfb < max_sfb; sfb++) { // use the sfb_offset[sfb] table from Annex B that matches the used sampling // frequency and the transform length related to the windows within group g sect_sfb_offset[g][sfb] = group_offset + sfb_offset[sfb] * num_win_in_group[g]; } group_offset += sfb_offset[max_sfb] * num_win_in_group[g]; }
4.3.6.2.7 get_max_sfb(g) – get max_sfb for group g
This helper function returns the max_sfb value for the window group g.
Pseudocode 5
get_max_sfb(g) { idx = 0; if (frame_len_base >= 1536 && (b_long_frame == 0) && (transf_length[0] != transf_length[1])) { num_windows_0 = (1 << (3-transf_length[0])); if (g >= window_to_group[num_windows_0]) { idx = 1; } } if ((b_side_limited == 1) || (b_dual_maxsfb == 1 && b_side_channel == 1)) { return max_sfb_side[idx]; } else { return max_sfb[idx]; } } // Note: b_side_channel=1 indicates the decoding of the side channel when // stereo_codec_mode == ASPX_ACPL_1.
4.3.6.3 asf_section_data - ASF section data
4.3.6.3.1 sect_cb[g][i] - section codebook
This element indicates the Huffman codebook to be used for section i in group g. The values 12 to 15 do not indicate a Huffman codebook and shall not be used.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)84
4.3.6.3.2 sect_len_incr - section length increment
This element is used for determining the section length of section i in group g. The length of the sect_len_incr field itself, n_sect_bits, depends on the transform length as shown in the pseudocode below.
Pseudocode 6
if (transf_length_g <= 2) { n_sect_bits = 3; } else { n_sect_bits = 5; }
4.3.6.4 asf_spectral_data - ASF spectral data
4.3.6.4.1 asf_qspec_hcw - Huffman coded quantized spectral lines
This element holds Huffman coded quantized spectral lines. Either two or four spectral lines, indicated by the used Huffman codebook dimension, are coded with one codeword.
4.3.6.4.2 huff_decode(hcb, hcw) - Huffman decoding
This function takes a Huffman codebook table and a Huffman codeword as input and returns the index of the Huffman codeword in the Huffman codebook table.
4.3.6.4.3 quad_sign_bits - quad sign bits
This element contains additional sign bits that are transmitted for each quantized spectral line that is different to zero when an unsigned Huffman codebook of dimension four is used for the quantized spectral lines.
4.3.6.4.4 pair_sign_bits - pair sign bits
This element contains additional sign bits that are transmitted for each quantized spectral line that is different to zero when an unsigned Huffman codebook of dimension two is used for the quantized spectral lines.
4.3.6.4.5 ext_code - extension code
This element is an extension code that is used in connection with Huffman codebook 11. For details, see clause 5.1.2.
4.3.6.5 asf_scalefac_data - ASF scale factor data
4.3.6.5.1 reference_scale_factor - reference scale factor
This element specifies the reference scale factor. The first scale factor used in the decoding process is equal to the reference scale factor. The remaining scale factor values are derived via delta decoding. See clause 6.2.6.4 for details.
4.3.6.5.2 asf_sf_hcw - Huffman coded scale factor delta
This element holds a Huffman coded delta value used for the dpcm_sf[g][sfb] determination. See clause 6.2.6.4 for details.
4.3.6.6 asf_snf_data - ASF spectral noise fill data
4.3.6.6.1 b_snf_data_exists - spectral noise fill data exists flag
If true, this Boolean indicates that spectral noise fill data shall follow in the bitstream.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)85
4.3.6.6.2 asf_snf_hcw - Huffman code spectral noise fill delta
This element holds a Huffman coded delta value used for the dpcm_snf[g][sfb] determination. See clause 5.1.4 for details.
4.3.7 Speech spectral frontend
4.3.7.1 ssf_data - Speech spectral frontend data
4.3.7.1.1 b_ssf_iframe - SSF I-frame flag
If true, this Boolean indicates that the first SSF granule in the ssf_data is an SSF-I-frame. The first SSF granule in the ssf_data is an SSF-I-frame if either b_iframe in the ac4_substream_info or b_ssf_iframe is true.
4.3.7.2 ssf_granule - speech spectral frontend granule
4.3.7.2.1 stride_flag - stride flag
This value of this element indicates the SSF coder mode and the number of SSF blocks per SSF granule, num_blocks, as shown in table 110. The stride_flag shall be set to 0 for SSF configurations that do not allow for a short stride mode as indicated by a value of 1 in the "Maximum number of SSF blocks" column in table 111 and table 112.
Table 110: SSF coder mode
stride_flag SSF coder mode Name num_blocks 0 long stride mode LONG_STRIDE 1 1 short stride mode SHORT_STRIDE 4
4.3.7.2.2 SSF configuration
The SSF configuration depends on the sampling frequency and the frame_rate_index as shown in table 111 and table 112.
Table 111: SSF configuration for 48 kHz sampling frequency
frame_rate_index Video frame rate
(fps)
Frame length (samples)
frame_length
SSF granule length (samples)
granule_length
Maximum number of SSF
blocks
Number of SSF granules
0 23,976 1 920 960 4 2 1 24 1 920 960 4 2 2 25 2 048 1 024 4 2 3 29,97 1 536 768 4 2 4 30 1 536 768 4 2 5 47,95 960 960 4 1 6 48 960 960 4 1 7 50 1 024 1 024 4 1 8 59,94 768 768 4 1 9 60 768 768 4 1 10 100 512 512 1 1 11 119,88 384 384 1 1 12 120 384 384 1 1 13 (23,44) 2 048 1 024 4 2 14+15 Reserved
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)86
Table 112: SSF configuration for 44,1 kHz sampling frequency
frame_rate_index Video frame rate
(fps)
Frame length (samples)
frame_length
SSF granule length (samples)
granule_length
Maximum number of SSF
blocks
Number of SSF
granules 0…12: reserved - - - - - 13 11 025/512 2 048 1 024 4 2 14+15: reserved - - - - -
The SSF block length in samples, n_mdct, is given by:
n_mdct =granule_length
num_blocks
4.3.7.2.3 num_bands_minus12 - number of SSF coded bands minus 12
This element indicates the number of SSF coded bands minus 12. To get the number of SSF coded bands, num_bands, a value of 12 needs to be added to num_bands_minus12.
4.3.7.2.4 predictor_presence_flag[b] - predictor presence flag for block b
If set, this bit indicates the existence of predictor parameters in the bitstream for block b.
4.3.7.2.5 delta_flag[b] - delta coding flag for block b
If set, this bit indicates the usage of the differential coding of the prediction lag for block b.
4.3.7.3 ssf_st_data - speech spectral frontend static data
4.3.7.3.1 env_curr_band0_bits - signal envelope index for band 0
This element indicates the signal envelope index for band 0. The envelope decoder expects this value in env_idx[0] .
4.3.7.3.2 env_startup_band0_bits - startup envelope index for band 0
This element indicates the extra startup envelope index for band 0.
4.3.7.3.3 gain_bits[b] - envelope gain bits for block b
This element indicates a gain index for block b. The gain index is derived by adding an offset value of -8 to the gain bits value: gain_idx[b] = gain_bits[b] - 8.
4.3.7.3.4 predictor_lag_delta_bits[b] - predictor lag delta for block b
This element indicates a predictor lag delta to be used for the predictor lag index calculation for block b. See clause 5.2.4.
4.3.7.3.5 predictor_lag_bits[b] - predictor lag index for block b
This element indicates the predictor lag index directly for block b.
4.3.7.3.6 variance_preserving_flag[b] - variance preserving flag for block b
If set, this bit indicates that variance preserving for block b shall be used.
4.3.7.3.7 alloc_offset_bits[block] - allocation offset bits for block b
This element indicates a value to be used for calculating an allocation offset for block b. See clause 5.2.5 for usage details.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)87
4.3.7.4 ssf_ac_data - speech spectral frontend arithmetic coded data
4.3.7.4.1 env_curr_ac_bits - arithmetic coded signal envelope indices
This element contains num_bands - 1 arithmetic coded signal envelope indices for band indices 1 to num_bands - 1. The arithmetic decoded values are stored in env_idx[band] , which is an input signal of the envelope decoder.
4.3.7.4.2 env_startup_ac_bits - arithmetic coded startup envelope indices
This element contains num_bands - 1 arithmetic coded extra startup envelope indices for band indices 1 to num_bands - 1. The startup envelope is decoded like the signal envelope and the decoded startup envelope is used as previous envelope env_prev[band] by the envelope decoder.
4.3.7.4.3 predictor_gain_ac_bits[b] - arithmetic coded predictor gain index for block b
This element contains one arithmetic coded predictor gain index for block b.
4.3.7.4.4 q_mdct_coefficients_ac_bits[b] - arithmetic coded quantized MDCT coefficients for block b
This element contains num_bins arithmetic coded quantized MDCT transform coefficients for block b.
4.3.7.5 SSF helper elements
The following helper elements are derived from the SSF bitstream elements:
start_bin[k] Vector indicating the coefficient index for the start of band k.
end_bin[k] Vector indicating the coefficient index for the end of band k.
num_bins Number of coded spectral coefficients.
The following pseudocode shows how to initialize the helper elements using num_bands, n_mdct, and the SSF bandwidths given in table C.1.
Pseudocode 7
// initialize start_bin[], stop_bin[] and num_bins //band_widths is a column from table C.1 selected by n_mdct MAX_NUM_BANDS = 19; start_bin[0] = 0; end_bin[0] = band_widths[0] - 1; for (i = 1; i < MAX_NUM_BANDS; i++) { start_bin[i] = start_bin[i-1] + band_widths[i-1]; end_bin[i] = start_bin[i] + band_widths[i] - 1; } num_bins = end_bin[num_bands - 1] + 1;
4.3.8 Stereo audio processing
4.3.8.1 chparam_info - stereo information
4.3.8.1.1 sap_mode - stereo audio processing mode
The value of this element indicates the stereo audio processing mode as shown in table 113.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)88
Table 113: Stereo audio processing mode
sap_mode SAP mode 0 No SAP 1 M/S processing in scale factor bands specified by ms_used[g][sfb] 2 M/S processing in all scale factor bands 3 Full SAP
4.3.8.1.2 ms_used - M/S coding used
If set, this bit indicates that M/S coding is used in scale factor band sfb in group g.
4.3.8.2 sap_data - stereo audio processing data
4.3.8.2.1 sap_coeff_all - SAP coding all scale factor bands flag
If set, this bit indicates that SAP coding is used in all scale factor bands. If not set, SAP coding is used in scale factor bands specified by sap_coeff_used[g][sfb].
4.3.8.2.2 sap_coeff_used - SAP coding used in scale factor band sfb
If set, this bit indicates that SAP coding is used in scale factor band sfb in group g.
4.3.8.2.3 delta_code_time - delta coding in time
This element is used to determine delta coding in time of the alpha_q values.
If this element is set and group g > 0, the alpha_q deltas apply to the alpha_q values from group g - 1.
If this element is not set or group g = 0, the alpha_q deltas apply to the alpha_q values from scale factor band sfb - 2.
See clause 5.3.1 for details.
4.3.8.2.4 sap_hcw - Huffman coded alpha_q delta
This element contains a Huffman coded delta value used for the alpha_q[g][sfb] determination. See clause 5.3.1 for details. The same Huffman codebook as for the ASF scale factor deltas is used.
4.3.9 Companding control
4.3.9.1 sync_flag
The sync_flag indicates the cross-channel synchronization according to table 114.
Table 114: sync_flag
sync_flag Description 0 Cross-channel synchronization disabled 1 Cross-channel synchronization enabled
4.3.9.2 b_compand_on
This flag is present per channel ch and indicates whether companding shall be used for a given channel ch individually. The values for b_compand_on are given in table 115. If b_compand_on is set to 0, the given channel ch shall either be unprocessed or the gain value shall be averaged, which is signalled via b_compand_avg.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)89
Table 115: b_compand_on
b_compand_on Description 0 Companding shall not be used for this channel individually (see also b_compand_avg) 1 Companding shall be used for this channel individually
4.3.9.3 b_compand_avg
The b_compand_avg flag indicates whether gains shall be averaged and held constant across a frame for all channels with b_compand_on ch == 0 according to table 116.
Table 116: b_compand_avg
b_compand_avg Description 0 Channels with b_compand_onch == 0 shall be unprocessed 1 Gain values for channels with b_compand_onch == 0 shall be averaged and held constant across a
frame
4.3.10 Advanced spectral extension - A-SPX
4.3.10.1 aspx_config - A-SPX configuration
4.3.10.1.1 aspx_quant_mode_env
This element is used as an initial value for the variable aspx_qmode_env[ch], which indicates the size of quantization steps used for the encoded signal envelopes, as shown in table 117. See clause 5.7.6.3.5 for details.
Table 117: aspx_quant_mode_env[ch]
aspx_quant_mode_env Quantization step size 0 1,5 dB 1 3,0 dB
4.3.10.1.2 aspx_start_freq - A-SPX start QMF subband
This element is an index into the scale factor subband group table, starting from the first subband and moving upwards in steps of 2 subbands. An aspx_start_freq of 1 hence points to subband 20 for the high-frequency resolution table. See clause 5.7.6.3.1.1 for more details.
4.3.10.1.3 aspx_stop_freq - A-SPX stop QMF subband
This element is an index into the scale factor subband group table, starting from the last subband and going downwards in steps of 2 subbands. An aspx_stop_freq of 2 hence points to subband 50 in the high-frequency resolution table. See clause 5.7.6.3.1.1 for more details.
4.3.10.1.4 aspx_master_freq_scale - A-SPX master frequency table scale
The value of this element indicates which of the two static subband group tables is to be used to generate the master subband group table, according to table 118.
Table 118: aspx_master_freq_scale
aspx_master_freq_scale Meaning 0 Low bit-rate scale factor table (sbg_template_lowres) 1 High bit-rate scale factor table (sbg_template_highres)
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)90
4.3.10.1.5 aspx_interpolation - A-SPX interpolation used
The value of this element determines how to estimate the energy of envelopes within an A-SPX interval. The estimation is performed by calculating the average of the squared complex subband samples over the time and frequency regions of the QMF time-frequency matrix. Possible values of aspx_interpolation are given in table 119.
See clause 5.7.6.4.2.1 for more details.
Table 119: aspx_interpolation
aspx_interpolation Meaning 0 Interpolation not used 1 Interpolation used
4.3.10.1.6 aspx_preflat - A-SPX pre-flattening used
The value of this element indicates whether spectral pre-flattening is used, according to table 120.
Table 120: aspx_preflat
aspx_preflat Meaning 0 Pre-flattening not used 1 Pre-flattening used
4.3.10.1.7 aspx_limiter - A-SPX limiter used
The value of this element indicates whether the limiter is turned on or off for a particular A-SPX interval, according to table 121.
Table 121: aspx_limiter
aspx_limiter Meaning 0 Limiter off 1 Limiter on
4.3.10.1.8 aspx_noise_sbg - A-SPX number of noise subband groups
This element is an input parameter to the function that calculates the number of noise subband groups, as detailed in clause 5.7.6.3.1.3.
4.3.10.1.9 aspx_num_env_bits_fixfix - A-SPX frame class FIXFIX bit count
This element is an input parameter to the aspx_framing bitstream parsing block described in clause 4.3.10.4. The value of this element indicates whether 1 or 2 bits are used for the transmission of the tmp_num_env bitstream parameter in the FIXFIX interval class. In the FIXFIX interval class the number of signal envelopes is derived from tmp_num_env as aspx_num_env[ch] = (1 << tmp_num_env). Hence, transmitting tmp_num_env using 1 bit only allows the use of either one or two signal envelopes, while 2 bits additionally allows the use of four uniformly spaced signal envelopes. Eight envelopes is prohibited by the A-SPX syntax. See table 127.
Table 122: aspx_num_env_bits_fixfix
aspx_num_env_bits_fixfix Meaning (FIXFIX class only) 0 tmp_num_env is transmitted using 1 bit 1 tmp_num_env is transmitted using 2 bits
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)91
4.3.10.1.10 aspx_freq_res_mode - A-SPX frequency resolution transmission mode
This element is an input parameter to the aspx_framing bitstream parsing block described in clause 4.3.10.4 and to the function that calculates the A-SPX interval borders, as outlined in clause 5.7.6.3.3.1.
Table 123: aspx_freq_res_mode
aspx_freq_res_mode Meaning 0 aspx_freq_res is signalled in aspx_framing 1 aspx_freq_res defaults to low resolution 2 aspx_freq_res defaults to predetermined values depending on the signal envelope duration 3 aspx_freq_res defaults to high resolution
4.3.10.2 aspx_data_1ch - A-SPX 1-channel data
4.3.10.2.1 aspx_xover_subband_offset - A-SPX crossover subband offset
This element is an index into the high-frequency resolution subband group table starting at the first subband and moving upward with a step of 1 subband.
4.3.10.3 aspx_data_2ch - A-SPX 2-channel data
4.3.10.3.1 aspx_xover_subband_offset - A-SPX crossover subband offset
This 3-bit element has been described in clause 4.3.10.2.1.
4.3.10.3.2 aspx_balance - A-SPX balance setting
The value of this element indicates whether the envelope and noise scale factors of the two channels have been coded separately or paired, according to table 124.
Table 124: aspx_balance
aspx_balance Coding of envelope and noise scale factors 0 Separate per-channel coding 1 Paired channel coding
4.3.10.4 aspx_framing - A-SPX framing
4.3.10.4.1 aspx_int_class - A-SPX interval class
The value of this element indicates the A-SPX interval class used in the bitstream as shown in table 125.
Table 125: aspx_int_class
aspx_int_class Interval class 0b0 FIXFIX 0b10 FIXVAR 0b110 VARFIX 0b111 VARVAR
The information transmitted in the bitstream depends on the indicated interval class, as follows:
FIXFIX:
• the number of uniformly spaced signal envelopes; and
• the frequency resolution (coarse or fine) for all envelopes.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)92
FIXVAR:
• the location of the trailing (right) interval border;
• the number of relative borders computed from the end of the interval;
• vectors containing relative borders associated with the trailing interval border;
• a pointer indicative of the signal envelope of transient characteristics if any; and
• the frequency resolution (coarse or fine) of each envelope.
VARFIX:
• the location of the leading (left) interval border;
• the number of relative borders computed from the beginning of the interval;
• vectors containing relative borders associated with the leading interval border;
• a pointer indicative of the signal envelope of transient characteristics if any; and
• the frequency resolution (coarse or fine) of each envelope.
VARVAR:
• the location of the leading (left) interval border;
• the number of relative borders computed from the beginning of the interval;
• vectors containing relative borders associated with the leading interval border;
• the location of the trailing (right) interval border;
• the number of relative borders computed from the end of the interval;
• vectors containing relative borders associated with the trailing interval border;
• a pointer indicative of the signal envelope of transient characteristics if any; and
• the frequency resolution (coarse or fine) of each envelope.
For the FIXFIX interval class, the signal envelopes are uniformly spaced within the A-SPX interval, while the FIXVAR, VARFIX and VARVAR interval classes do not generally have uniformly spaced signal envelopes.
One A-SPX interval can either be one or two noise envelopes. If the number of signal envelopes is 1, the interval contains one noise envelope, and if the number of signal envelopes is greater than 1, the interval contains two noise envelopes. In the latter case, the placement of the border between the two noise envelopes is a function of the variable aspx_tsg_ptr and the frame interval class aspx_int_class.
4.3.10.4.2 tmp_num_env - temporary variable
This element is a temporary variable that is used to calculate the variable aspx_num_env[ch] described in clause 4.3.10.4.11.
4.3.10.4.3 aspx_freq_res[ch][env] - frequency resolution
The value of this element indicates the frequency resolution, low or high, for each channel and signal envelope, as shown in table 126.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)93
Table 126: aspx_freq_res
aspx_freq_res Meaning 0 Low-frequency resolution 1 High-frequency resolution
4.3.10.4.4 aspx_var_bord_left[ch] - leading VAR interval envelope border
This 2-bit element indicates the position of the leading variable time slot group border for signal envelopes for the interval classes VARFIX and VARVAR.
4.3.10.4.5 aspx_var_bord_right[ch] - trailing VAR interval envelope border - 2-bits
This element indicates the position of the trailing variable time slot group border for signal envelopes for interval classes FIXVAR and VARVAR.
4.3.10.4.6 aspx_num_rel_left[ch] - relative envelope border - 1 or 2 bits
This 1- or 2-bit element indicates the number of relative time slot group borders for signal envelopes, starting from aspx_var_bord_left. For all frame interval classes, 2 bits are read if num_aspx_timeslots is greater than 8; otherwise, 1 bit is read.
4.3.10.4.7 aspx_num_rel_right[ch] - relative envelope border - 1 or 2 bits
This element indicates the number of relative time slot group borders for signal envelopes, starting from aspx_var_bord_right. For all frame interval classes, 2 bits are read if num_aspx_timeslots is greater than 8; otherwise, 1 bit is read.
4.3.10.4.8 aspx_rel_bord_left[ch][rel] - leading relative envelope borders
Array of variables indicating the respective offset of each border to the border preceding it, starting left. The position of the first, leftmost border (which has no preceding border) is calculated relative to the position indicated by aspx_var_bord_left[ch], i.e. the first border's position is calculated by adding the first border's offset to aspx_var_bord_left[ch].
For all frame interval classes, 2 bits are read if num_aspx_timeslots is greater than 8; otherwise, 1 bit is read.
4.3.10.4.9 aspx_rel_bord_right[ch][rel] - trailing relative envelope borders
Array of variables indicating the respective offset of each border to the border preceding it, starting right. The position of the first, rightmost border (which has no preceding border) is calculated relative to the position indicated by aspx_var_bord_right[ch], i.e. the first border's position is calculated by subtracting the first border's offset from aspx_var_bord_right[ch].
For all frame interval classes, 2 bits are read if num_aspx_timeslots is greater than 8; otherwise, 1 bit is read.
4.3.10.4.10 aspx_tsg_ptr[ch] - pointer to envelope border
This variable length element is a pointer to a specific A-SPX time slot group border, used to determine the position of the A-SPX time slot for placing sinusoids and noise envelope borders.
See clause 5.7.6.3.3.1 and clause 5.7.6.4.2.1 for more details.
4.3.10.4.11 A-SPX framing helper variables
The following helper elements are derived from the A-SPX framing elements:
aspx_num_env[ch] Variable indicating the number of signal envelopes per channel.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)94
aspx_num_noise[ch] Variable indicating the number of noise envelopes per channel.
In a valid bitstream, the number of signal envelopes in an A-SPX interval satisfies the conditions described in table 127.
Table 127: Limit for the number aspx_num_env of envelopes
aspx_int_class max(aspx_num_env) FIXFIX 4 FIXVAR, VARFIX 5 VARVAR 5
4.3.10.5 aspx_delta_dir - A-SPX delta coding direction
4.3.10.5.1 aspx_sig_delta_dir[ch][env] - A-SPX delta coding for signal envelopes
The value of this element indicates whether the Huffman values for the signal envelopes are coded in time or frequency direction as shown in table 128.
Table 128: aspx_sig_delta_dir
aspx_sig_delta_dir Meaning 0 Delta coding in frequency direction 1 Delta coding in time direction
4.3.10.5.2 aspx_noise_delta_dir[ch][env] - A-SPX delta coding for noise envelopes
The value of this element indicates whether the Huffman values for the noise envelopes are coded in time or frequency direction as shown in table 129.
Table 129: aspx_noise_delta_dir
aspx_noise_delta_dir Meaning 0 Delta coding in frequency direction 1 Delta coding in time direction
4.3.10.6 aspx_hfgen_iwc_1ch - A-SPX 1-channel HF generation and interleaved waveform coding
4.3.10.6.1 aspx_tna_mode[n] - A-SPX subband tonal to noise ratio adjustment mode
The 2-bit subband tonal to noise ratio adjustment mode signalled for an A-SPX interval is one of the four values shown in table 130.
Table 130: aspx_tna_mode
aspx_tna_mode Meaning 0 None 1 Light 2 Moderate 3 Heavy
The amount of tonal to noise ratio adjustment is proportional to a value calculated based on the values of aspx_tna_mode for the current A-SPX interval and the previous A-SPX interval.
For more detail on the subband tonal to noise ratio adjustment, see clause 5.7.6.4.1.3.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)95
4.3.10.6.2 aspx_add_harmonic[n] - A-SPX add harmonics
This bit indicates the insertion of a sinusoid in a QMF subband where there was none present in the previous A-SPX interval, according to table 131.
Table 131: aspx_add_harmonic
aspx_add_harmonic Meaning 0 Do not add sinusoid 1 Add sinusoid
4.3.10.6.3 aspx_fic_present - A-SPX frequency interleaved coding present
This bit indicates the presence of frequency interleaved coding, according to table 132.
Table 132: aspx_fic_present
aspx_fic_present Meaning 0 Frequency interleaved coding present 1 Frequency interleaved coding present
4.3.10.6.4 aspx_fic_used_in_sfb[n] - A-SPX frequency interleaved coding used in subband group
The value of this element indicates whether frequency interleaved coding is used in a particular subband group, as shown in table 133.
Table 133: aspx_fic_used_in_sfb
aspx_fic_used_in_sfb Meaning 0 Frequency interleaved coding not used in current subband group 1 Frequency interleaved coding used in current subband group
4.3.10.6.5 aspx_tic_present - A-SPX time interleaved coding present
The value of this element indicates whether time interleaved coding is present, as shown in table 134.
Table 134: aspx_tic_present
aspx_tic_present Meaning 0 Time interleaved coding not present 1 Time interleaved coding present
4.3.10.6.6 aspx_tic_used_in_slot[n] - A-SPX time interleaved coding used in slot
The value of this element indicates whether time interleaved coding is used in an adjacent pair of QMF slots, as shown in table 135.
Table 135: aspx_tic_used_in_slot
aspx_tic_used_in_slot Meaning 0 Time interleaved coding not used in current slot pair 1 Time interleaved coding used in current slot pair
4.3.10.6.7 aspx_ah_present - A-SPX add harmonics present
If set, this bit indicates that aspx_add_harmonic[n] data shall follow in the bitstream.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)96
4.3.10.7 aspx_hfgen_iwc_2ch - A-SPX 2-channel HF generation and interleaved waveform coding
4.3.10.7.1 aspx_tna_mode[ch][n] - A-SPX subband tonal to noise adjustment mode
This element is described in clause 4.3.10.6.1, and takes an additional channel index.
4.3.10.7.2 aspx_add_harmonic[ch][n] - A-SPX add harmonics
This element is described in clause 4.3.10.6.2, and takes an additional channel index.
4.3.10.7.3 aspx_fic_present - A-SPX frequency interleaved coding present - 1 bit
This element is described in clause 4.3.10.6.3.
4.3.10.7.4 aspx_fic_left - A-SPX frequency interleaved coding in left channel
The value of this element indicates whether frequency interleaved coding is present in the left channel, as shown in table 136.
Table 136: aspx_fic_left
aspx_fic_left Meaning 0 Frequency interleaved coding not present in left channel 1 Frequency interleaved coding present in left channel
4.3.10.7.5 aspx_fic_right - A-SPX frequency interleaved coding in right channel
The value of this element indicates whether frequency interleaved coding is present in the right channel, according to table 137.
Table 137: aspx_fic_right
aspx_fic_right Meaning 0 Frequency interleaved coding not present in right channel 1 Frequency interleaved coding present in right channel
4.3.10.7.6 aspx_fic_used_in_sfb[ch][n] - A-SPX frequency interleaved coding used in subband group
This element is described in clause 4.3.10.6.4, and takes an additional channel index.
4.3.10.7.7 aspx_tic_present - A-SPX time interleaved coding present
This element is described in clause 4.3.10.6.5.
4.3.10.7.8 aspx_tic_copy - A-SPX time interleaved coding copy data
The value of this element indicates whether the time interleaved data is copied to both left and right channels, as shown in table 138.
Table 138: aspx_tic_copy
aspx_tic_copy Meaning 0 Time interleaved coding not copied 1 Time interleaved coding copied
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)97
4.3.10.7.9 aspx_tic_left - A-SPX time interleaved coding in left channel
The value of this element indicates whether time interleaved coding is present in the left channel, as shown in table 139.
Table 139: aspx_tic_left
aspx_tic_left Meaning 0 Time interleaved coding not present in left channel 1 Time interleaved coding present in left channel
4.3.10.7.10 aspx_tic_right - A-SPX time interleaved coding in right channel
The value of this element indicates whether time interleaved coding is present in the right channel, as shown in table 140.
Table 140: aspx_tic_right
aspx_tic_right Meaning 0 Time interleaved coding not present in right channel 1 Time interleaved coding present in right channel
4.3.10.7.11 aspx_tic_used_in_slot[n] - A-SPX time interleaved coding used in slots
This element is described in clause 4.3.10.6.6.
4.3.10.7.12 aspx_ah_left - A-SPX add harmonics in left channel
If set, this bit indicates that aspx_add_harmonic[0][n] data shall follow in the bitstream.
4.3.10.7.13 aspx_ah_right - A-SPX add harmonics in right channel
If set, this bit indicates that aspx_add_harmonic[1][n] data shall follow in the bitstream.
4.3.10.8 Functions for Huffman coding
4.3.10.8.0 aspx_ec_data - A-SPX Huffman data
This function returns the Huffman decoded signal and noise envelope scale factors from the bitstream. It takes the following arguments:
data_type Variable indicating the whether to decode a signal or a noise envelope.
num_env Variable indicating the number of envelopes to decode.
freq_res Variable indicating the frequency resolution of signal envelopes.
quant_mode Variable indicating the quantization mode for signal envelopes.
stereo_mode Variable indicating the whether to decode two channels separately or paired.
direction Variable indicating the direction of the delta coding.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)98
4.3.10.8.1 aspx_huff_data - Huffman decoding for A-SPX values
This function returns an array of scale factors of a decoded signal or noise envelope.
It takes the following arguments:
data_type Variable indicating the whether to decode a signal or a noise envelope.
sbg Variable indicating the number of subband groups in the current envelope.
qm Variable indicating the quantization mode for signal envelopes.
sm Variable indicating the whether to decode two channels separately or paired.
dir Variable indicating the direction of the delta coding.
4.3.10.8.2 aspx_hcw - A-SPX Huffman code word
This element indicates the Huffman code word used for Huffman decoding.
4.3.10.8.3 huff_decode_diff(hcb, hcw) - Huffman decoding for differences
This function takes a Huffman codebook table and a Huffman codeword as input and returns the index of the Huffman codeword in the Huffman codebook table subtracted by the Huffman codebook offset cb_off, which is specified together with the Huffman codebook table.
4.3.11 Advanced coupling - A-CPL
4.3.11.1 acpl_config_1ch - A-CPL 1-channel configuration
4.3.11.1.1 acpl_1ch_mode - A-CPL tool 1-channel mode
This helper element indicates the advanced coupling (A-CPL) tool 1-channel mode as shown in table 141.
Table 141: acpl_1ch_mode
acpl_1ch_mode Meaning 0 Full 1 Partial
4.3.11.1.2 acpl_num_param_bands_id - A-CPL number of parameter bands
The value of this element indicates the number of parameter bands, as shown in table 142.
A parameter band is a grouping of QMF subbands. Advanced coupling parameters are transmitted per parameter band.
Table 142: acpl_num_param_bands_id
acpl_num_param_bands_id acpl_num_param_bands 0 15 1 12 2 9 3 7
4.3.11.1.3 acpl_quant_mode - A-CPL quantization mode
The value of this element indicates whether coarse or fine quantization is used.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)99
Table 143: acpl_quant_mode
acpl_quant_mode Meaning 0 Fine 1 Coarse
4.3.11.1.4 acpl_qmf_band_minus1 - A-CPL QMF band minus 1
Adding a value of 1 to this 3-bit code determines the value of the helper variable acpl_qmf_band. acpl_qmf_band indicates a QMF subband below which the signal is mid-side coded, and above or at which, the signal is coded using advanced coupling.
4.3.11.2 acpl_config_2ch - A-CPL 2-channel configuration
4.3.11.2.1 acpl_num_param_bands_id - A-CPL number of parameter bands
This 2-bit bitstream element is described in clause 4.3.11.1.2.
4.3.11.2.2 acpl_quant_mode_0 - A-CPL quantization mode 0
This bit indicates coarse or fine quantization for acpl_alpha and acpl_beta coefficients. The possible values of this bitstream element are shown in clause 4.3.11.1.3.
4.3.11.2.3 acpl_quant _mode_1 - A-CPL quantization mode 1
This bit indicates coarse or fine quantization for acpl_gamma coefficients. The possible values of this bitstream element are shown in clause 4.3.11.1.3.
4.3.11.3 acpl_data_1ch - A-CPL 1-channel data
The acpl_alpha1 and acpl_beta1 identifiers are used for Huffman table dequantization.
4.3.11.4 acpl_data_2ch - A-CPL 2-channel data
The identifiers used for Huffman table dequantization are: acpl_alpha1, acpl_alpha2, acpl_beta1, acpl_beta2, acpl_beta3, acpl_gamma1, acpl_gamma2, acpl_gamma3, acpl_gamma4, acpl_gamma5, and acpl_gamma6.
4.3.11.5 acpl_framing_data - A-CPL framing data
4.3.11.5.1 acpl_interpolation_type - A-CPL interpolation type
This bit indicates the type of interpolation used.
Table 144: acpl_interpolation_type
acpl_interpolation_type Meaning 0 Smooth A-CPL interpolation 1 Steep A-CPL interpolation
4.3.11.5.2 acpl_num_param_sets_cod - A-CPL number of parameter sets per frame
This bit indicates the value acpl_num_param_sets, which describes how many parameter sets per frame are transmitted in the bitstream.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)100
Table 145: acpl_num_param_sets
acpl_num_param_sets_cod acpl_num_param_sets 0 1 1 2
4.3.11.5.3 acpl_param_timeslot - A-CPL parameter change at QMF time slot
When steep interpolation is used, this 5-bit element indicates the QMF time slot (0 - 31) at which the parameter set values change.
4.3.11.6 acpl_huff_data - A-CPL Huffman data
4.3.11.6.1 acpl_hcw - A-CPL Huffman code word
This variable length element indicates the Huffman code word used for Huffman decoding. The following pseudocode describes how to choose the correct Huffman table for decoding the bitstream elements.
Pseudocode 8
get_acpl_hcb(data_type, quant_mode, hcb_type) { // data_type = {ALPHA, BETA, BETA3, GAMMA} // quant_mode = {COARSE, FINE) // hcb_type = {F0, DF, DT} acpl_hcb = ACPL_HCB_<data_type>_<quant_mode>_<hcb_type>; // the line above expands using the inputs data_type, quant_mode and hcb_type // The 24 A-CPL Huffman codebooks are given in table A.34 to table A.57 // and are named according to the schema outlined above. return acpl_hcb; }
4.3.12 Basic and extended metadata
4.3.12.1 metadata - metadata
4.3.12.1.1 tools_metadata_size_value - size of tools metadata
This element generates tools_metadata_size, indicating the size in bits of the metadata related to the dynamic range control (DRC) and dialogue enhancement tools. It is a 7-bit element that can be extended by variable_bits().
4.3.12.1.2 b_more_bits - more bits flag
If true, this Boolean indicates the presence of additional variable_bits to be used for the tools_metadata_size determination.
4.3.12.1.3 b_emdf_payloads_substream - EMDF payloads substream flag
If true, this Boolean indicates the presence of an emdf_payloads_substream() element.
4.3.12.2 basic_metadata - basic metadata
4.3.12.2.1 dialnorm_bits - input reference level
This 7-bit element specifies the dialnorm in dBFS, in steps of ¼ dBFS. The dialnorm value indicates the input reference level.
dialnorm = -0,25 × ��������_��[dBFS]
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)101
4.3.12.2.2 b_more_basic_metadata - more basic metadata flag
If true, this Boolean indicates that more basic metadata is available.
4.3.12.2.3 b_further_loudness_info - additional loudness information flag
If true, this Boolean indicates that additional loudness information is present in a further_loudness_info() element.
4.3.12.2.4 b_prev_dmx_info - previous downmix information flag
If true, this Boolean indicates that downmix information of the signal before encoding is present.
4.3.12.2.5 pre_dmixtyp_2ch - previous downmix to 2 channels type
This 3-bit elementindicates what downmix was performed before encoding into two channels as shown in table 146.
Table 146: pre_dmixtyp_2ch
pre_dmixtyp_2ch Downmix equation 0 Unknown 1 Lo = L + (-3 dB) × C + (-3 dB) × Ls
Ro = R + (-3 dB) × C + (-3 dB) × Rs 2 Lt = L + (-3 dB) × C - (-3 dB) × Ls - (-3 dB) × Rs
Rt = R + (-3 dB) × C + (-3 dB) × Ls + (-3 dB) × Rs 3 Lt = L + (-3 dB) × C - (-1,2 dB) × Ls - (-6,2 dB) × Rs
Rt = R + (-3 dB) × C + (-1,2 dB) × Rs + (-6,2 dB) × Ls 4…7 Reserved
4.3.12.2.6 phase90_info_2ch - phase 90 in 2 channels info
This 2-bit element indicates whether a 90° phase filtering was done before downmixing and encoding into 2 channels as shown in table 147.
Table 147: phase90_info_2ch
phase90_info_2ch Meaning 0 Not indicated 1 Reserved 2 Surrounds have undergone a 90° phase shift before downmixing and encoding 3 Surrounds have not undergone a 90° phase shift before downmixing and encoding
4.3.12.2.7 b_dmx_coeff - downmix coefficients present flag
If true, this Boolean indicates that downmix coefficients are present.
4.3.12.2.8 loro_center_mixgain - LoRo Centre mix gain
This 3-bit element indicates the gain to be used for LoRo downmixing the Centre channel of a five-channel or seven-channel signal into two channels as shown in table 148.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)102
Table 148: loro_{center,surround}_mixgain
loro_center_mixgain loro_surround_mixgain
LoRo mixgain
000 1,414 (+3,0 dB) 001 1,189 (+1,5 dB) 010 1,000 (0,0 dB) 011 0,841 (-1,5 dB) 100 0,707 (-3,0 dB) 101 0,595 (-4,5 dB) 110 0,500 (-6,0 dB) 111 0,000 (-inf dB) When no mixgains have been transmitted, a downmix gain of -3,0 dB shall be used.
4.3.12.2.9 loro_surround_mixgain - LoRo surround mix gain
This 3-bit element indicates the gain to be used for LoRo downmixing the surround channels of a five-channel or seven-channel channel signal into two channels as shown in table 148.
4.3.12.2.10 b_loro_dmx_loud_corr - LoRo downmix loudness correction flag
If true, this Boolean indicates that LoRo downmix loudness correction data is present.
4.3.12.2.11 loro_dmx_loud_corr - LoRo downmix loudness correction
This 5-bit element indicates a LoRo downmix loudness correction. This correction factor is used to calibrate downmix loudness to the original program loudness:
loro_corr_gain = (15 − loro_dmx_loud_corr) 2[dB�]⁄
A value of 31 is reserved, and if present, it indicates a gain of 0 [dB].
4.3.12.2.12 b_ltrt_mixinfo - LtRt downmix information present
If true, this Boolean indicates that LtRt downmix coefficients are present. If false, the LtRt downmix coefficients are the same as the LoRo downmix coefficients.
4.3.12.2.13 ltrt_center_mixgain - LtRt Centre mix gain
This 3-bit element indicates the gain to be used for LtRt downmixing the Centre channel of a five-channel or seven-channel signal into two channels as shown in table 148.
4.3.12.2.14 ltrt_surround_mixgain - LtRt surround mix gain
This 3-bit element indicates the gain to be used for LtRt downmixing the surround channels of a five-channel or seven-channel signal into two channels as shown in table 148.
4.3.12.2.15 b_ltrt_dmx_loud_corr - LtRt downmix loudness correction flag
If true, this Boolean indicates that LtRt downmix loudness correction data is present.
4.3.12.2.16 ltrt_dmx_loud_corr - LtRt downmix loudness correction
This 5-bit element indicates a LtRt downmix loudness correction. This correction factor is used to calibrate downmix loudness to the original program loudness
ltrt_corr_gain = (15 − ltrt_dmx_loud_corr) 2⁄ [dB�]
A value of 31 is reserved, and if present, it indicates a gain of 0 [dB].
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)103
4.3.12.2.17 b_lfe_mixinfo - LFE downmix information present
If true, this Boolean indicates that an LFE mix gain is present.
4.3.12.2.18 lfe_mixgain - LFE mix gain
This 5-bit element indicates the gain lfe_mg to be used for mixing the LFE channel to L and R when downmixing from 5.1 into two channels:
lfe_mg [dB] = 5,5 − lfe_mixgain
The lfe_mg range is -25,5 dB … +5,5 dB.
NOTE: After startup or after a detected splice, a value of -∞ dB may be used for lfe_mg until an AC-4 frame with b_lfe_mixinfo = 1 is received.
4.3.12.2.19 preferred_dmx_method - preferred downmix method
This 2-bit element indicates the preferred way of downmixing from five channels to two channels as shown in table 149.
Table 149: preferred_dmx_method
preferred_dmx_method Used coefficients 0 Not indicated 1 loro 2 ltrt 3 ltrt
4.3.12.2.20 b_predmxtyp_5ch - previous downmix to five channels flag
If true, this Boolean indicates that downmix information of the signal before encoding is present.
4.3.12.2.21 pre_dmixtyp_5ch - previous downmix to five channels type
This 3-bit element indicates what downmix was performed before encoding into five channels as shown in table 150.
Table 150: pre_dmixtyp_5ch
pre_dmixtyp_5ch Downmix equation 0 L = L
R = R C = C
LFE = LFE Ls' = Ls + (-3 dB) × Cs Rs' = Rs + (-3 dB) × Cs
1 L = L R = R C = C
LFE = LFE Ls' = (-3 dB) × Ls + (-3 dB) × Lrs Rs' = (-3 dB) × Rs + (-3 dB) × Rrs
2 L = L R = R C = C
LFE = LFE Ls' = Ls + (-1,2 dB) × Lrs + (-6,2 dB) × Rrs Rs' = Rs + (-1,2 dB) × Rrs + (-6,2 dB) × Lrs
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)104
pre_dmixtyp_5ch Downmix equation 3 L = L
R = R C = C
LFE = LFE Ls' = Ls - (-1,2 dB) × Lvh - (-6,2 dB) × Rvh
Rs' = Rs + (-1,2 dB) × Rvh + (-6,2 dB) × Lvh 4…7 Reserved
4.3.12.2.22 b_preupmixtyp_5ch - previous upmix to five channels flag
If true, this Boolean indicates that upmix information of the signal before encoding is present.
4.3.12.2.23 pre_upmixtyp_5ch - previous upmix to five channels type
This 4-bit element indicates what upmix from 2 channels was performed before encoding into 5 channels as shown in table 151.
Table 151: pre_upmixtyp_5ch
pre_upmixtyp_5ch Upmix type 0 Dolby® Pro Logic® 1 Dolby® Pro Logic® II Movie Mode 2 Dolby® Pro Logic® II Music Mode 3 Dolby® Professional Upmixer 4…15 Reserved
4.3.12.2.24 b_upmixtyp_7ch - previous upmix to seven channels flag
If true, this Boolean indicates that upmix information of the signal before encoding is present.
4.3.12.2.25 pre_upmixtyp_3_4 - previous upmix to seven channels type
This 2-bit elementindicates what upmix from five channels was performed before encoding into seven channels (3/4/0) as shown in table 152.
Table 152: pre_upmixtyp_3_4
pre_upmixtyp_3_4 Upmix type 0 Dolby® Pro Logic® IIx Movie Mode 1 Dolby® Pro Logic® IIx Music Mode 2…3 Reserved
4.3.12.2.26 pre_upmixtyp_3_2_2 - previous upmix to seven channels type
This single-bit element indicates what upmix from five channels was performed before encoding into seven channels (3/2/2) as shown in table 153.
Table 153: pre_upmixtyp_3_2_2
pre_upmixtyp_3_2_2 Upmix type 0 Dolby® Pro Logic® IIz Height 1 Reserved
4.3.12.2.27 phase90_info_mc - phase 90 in multi-channel information
This 2-bit element indicates whether a 90° phase filtering was done before encoding as shown in table 154.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)105
Table 154: phase90_info_mc
phase90_info_mc Semantics 0 Not indicated 1 Surrounds have undergone a 90° phase shift before encoding 2 Surrounds have not undergone a 90° phase shift before encoding 3 Reserved
4.3.12.2.28 b_surround_attenuation_known - surround attenuation known flag
This bit should be ignored by the decoder.
4.3.12.2.29 b_lfe_attenuation_known - LFE attenuation known flag
This bit should be ignored by the decoder.
4.3.12.2.30 b_dc_blocking - DC blocking flag
If true, this Boolean indicates that DC blocking information is present.
4.3.12.2.31 dc_block_on - DC blocking
This bit, if set, indicates that the signal was DC filtered before encoding. If this bit is not set, the signal was not DC filtered before encoding.
4.3.12.3 further_loudness_info - additional loudness information
4.3.12.3.1 loudness_version - loudness version
This elementindicates the version of the loudness payload. It is a 2-bit element that is extendable via extended_loudness_version. For loudness data payloads that conform to the present document, the loudness_version field shall be set to 00, and thus the extended_loudness_version field shall not be present in the payload.
4.3.12.3.2 extended_loudness_version - loudness version extension
This 4-bit element, which is only present if loudness_version = 3, holds an extension to the loudness_version.
4.3.12.3.3 loud_prac_type - loudness practice type
This 4-bit element indicates which recommended practice for programme loudness measurement has been followed to generate the programme loudness of the current audio substream as shown in table 155.
Table 155: loud_prac_type
loud_prac_type Semantics 0000 Loudness regulation compliance not indicated 0001 Programme loudness measured according to ATSC A/85 [i.5] 0010 Programme loudness measured according to EBU R128 [i.6] 0011 Programme loudness measured according to ARIB TR-B32 [i.7] 0100 Programme loudness measured according to FreeTV OP-59 [i.8] 0101…1101 Reserved 1110 Manual 1111 Consumer leveller
4.3.12.3.4 b_loudcorr_dialgate - loudness correction dialogue gating flag
If true, this Boolean indicates whether the loudness of the programme has been corrected using dialogue gating.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)106
4.3.12.3.5 dialgate_prac_type - dialogue gating practice type
This 3-bit element indicates which dialogue gating method has been utilized for loudness correction with current audio substream as shown in table 156. The dialgate_prac_type parameter is only indicated if the b_loudcorr_dialgate parameter preceding in the bitstream is true.
Table 156: Dialogue gating method
dialgate_prac_type Meaning 000 Not indicated 001 Automated Centre or Left+Right channel(s) 010 Automated Left, Centre, and/or Right channel(s) 011 Manual selection 100…111 Reserved NOTE 1: dialgate_prac_type methods 001 and 010 are described in [i.9] and a value of 011 indicates that a manual
process was utilized to select representative portions of the program containing dialogue for measurement. NOTE 2: For dialgate_prac_type method 001, dialogue gating is applied to the Centre channel of a multichannel
program or to the power sum of Left and Right channels in a stereo program. NOTE 3: For dialgate_prac_type method 010, dialogue gating is applied individually to all front (main) channels of
the program.
4.3.12.3.6 b_loudcorr_type - loudness correction type
This Boolean indicates the type of loudness correction applied to the programme:
• If true, the loudness has been corrected using a real-time loudness measurement.
• If false, the loudness has been corrected with an infinite look-ahead (file-based) loudness correction process.
4.3.12.3.7 b_loudrelgat - loudness value relative gated flag
If true, this Boolean indicates that a loudrelgat field is present in the payload.
4.3.12.3.8 loudrelgat - loudness value relative gated
This 11-bit element indicates the integrated loudness of the audio programme, measured according to Recommendation ITU-R BS.1770 [i.3] and without any gain adjustments due to dialogue normalization and dynamic range compression being applied.
The loudrelgat parameter is encoded according to:
loudrelgat = �loudness_value_relative_gated×10+0,5� + 1024
and decoded according to: ��������_�����_�������_���� = (loudrelgat − 1024)/10[LUFS] 4.3.12.3.9 b_loudspchgat - loudness value speech gated flag
If true, this Boolean indicates that a loudspchgat and a dialgate_prac_type field is present in the payload.
4.3.12.3.10 loudspchgat - loudness value speech gated
This 11-bit element indicates the integrated dialogue-based loudness of the entire audio programme, measured according to formula (2) of Recommendation ITU-R BS.1770 [i.3] with dialogue gating. The loudspchgat value represents the dialogue-based loudness without any gain adjustments due to dialogue normalization and dynamic range compression being applied.
The loudspchgat parameter is encoded according to:
loudspchgat = �loudness_value_speech_gated×10+0,5� + 1024
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)107
and decoded according to: ��������_�����_�����ℎ_���� = (loudspchgat − 1024)/10[LUFS] 4.3.12.3.11 dialgate_prac_type - dialogue gating practice type
This 3-bit element indicates which dialogue gating method has been utilized to generate the loudspchgat parameter value for the current audio substream as shown in table 156. The dialgate_prac_type parameter is only indicated if the b_loudspchgat parameter preceding in the bitstream is true.
4.3.12.3.12 b_loudstrm3s - loudness values short term 3s flag
If true, this Boolean indicates that a loudstrm3s field is present in the payload.
4.3.12.3.13 loudstrm3s - loudness values short term 3s
This 11-bit element indicates the loudness of the preceding 3 seconds of the audio programme, measured according to Recommendation ITU-R BS.1771 [i.4] and without any gain adjustments due to dialogue normalization and dynamic range compression being applied.
The loudstrm3s parameter is encoded according to:
loudstrm3s = �loudness_value_short_term_3s×10+0,5� + 1024
and decoded according to: ��������_�����_�ℎ��_���_3� = (loudstrm3s − 1024)/10[LUFS] 4.3.12.3.14 b_max_loudstrm3s - maximum loudness value short term 3s flag
If true, this Boolean indicates that a max_loudstrm3s field follows in the bitstream.
4.3.12.3.15 max_loudstrm3s - maximum loudness value short term 3s
This 11-bit element indicates the maximum short-term loudness of the audio programme without any gain adjustments due to dialogue normalization and dynamic range compression being applied, measured according to Recommendation ITU-R BS.1771 [i.4].
The max_loudstrm3s parameter is encoded according to:
max_loudstrm3s = �max_loudness_value_short_term_3s×10+0,5� + 1024
and decoded according to: ���_��������_�����_�ℎ��_���_3� = (max_loudstrm3s − 1024)/10[LUFS] 4.3.12.3.16 b_truepk - true peak flag
If true, this Boolean indicates that a truepk field follows in the bitstream.
4.3.12.3.17 truepk - true peak
This 11-bit element indicates the true peak sample value of the programme measured since the previous value was sent in the bitstream. The truepk value is measured according to Annex 2 of Recommendation ITU-R BS.1770 [i.3].
The truepk parameter is encoded according to:
truepk = �true_peak_value×10+0,5� + 1024
and decoded according to: ���_����_����� = (truepk − 1024)/10[dBTP]
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)108
4.3.12.3.18 b_max_truepk - maximum true peak flag
If true, this Boolean indicates that a max_truepk field follows in the bitstream.
4.3.12.3.19 max_truepk - maximum true peak
This 11-bit element indicates the maximum true peak sample value of the programme measured according to Annex 2 of Recommendation ITU-R BS.1770 [i.3].
The max_truepk parameter is encoded according to:
max_truepk = �max_true_peak_value×10+0,5� + 1024
and decoded according to: ���_���_����_����� = (max_truepk − 1024)/10[dBTP] 4.3.12.3.20 b_prgmbndy - programme boundary flag
If true, this Boolean indicates that programme boundary data is present in the payload.
4.3.12.3.21 prgmbndy_bit - programme boundary bit
This bit, which can be present multiple times, is used for the determination of the programme boundary value prgmbndy. The value of the variable prgmbndy shall be equal to the number of frames between the current frame and the frame that contains the boundary between two different programmes. This data may be used to determine when to begin and end the measurement of the loudness parameters specified in this payload. An encoder shall restrict the transmitted value to the range [2; 512].
4.3.12.3.22 b_end_or_start - programme boundary end or start flag
If true, this Boolean indicates that the value of the prgmbndy variable indicates the number of frames between the current frame and the upcoming frame that contains the next programme boundary.
If false, this Boolean indicates that the value of the prgmbndy variable indicates the number of frames between the current frame and the past frame that contained the previous programme boundary.
4.3.12.3.23 b_prgmbndy_offset - programme boundary offset flag
If true, this Boolean indicates that a prgmbndy_offset field is present in the payload.
4.3.12.3.24 prgmbndy_offset - programme boundary offset
This 11-bit element indicates the offset in audio samples from the first sample of the frame indicated by the prgmbndy variable to the actual audio sample in that frame that corresponds to the programme boundary.
4.3.12.3.25 b_lra - loudness range flag
If true, this Boolean indicates that an lra field follows in the bitstream.
4.3.12.3.26 lra - loudness range
This 10-bit element indicates the loudness range of the programme as specified in EBU Tech Document 3342 [i.10].
The lra parameter is encoded according to:
lra = �loudness_range×10+0,5� and decoded according to: ��������_����� = lra/10[LU]
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)109
4.3.12.3.27 lra_prac_type - loudness range measurement practice type
This 3-bit element field indicates which method has been utilized to compute the loudness range with the current audio substream as shown in table 157.
Table 157: lra_prac_type
lra_prac_type Semantics 000 Loudness Range as per EBU Tech 3342 [i.10] V1 001 Loudness Range as per EBU Tech 3342 [i.10] V2 010…111 Reserved
4.3.12.3.28 b_loudmntry - momentary loudness flag
If true, this Boolean indicates that a loudmntry field follows in the bitstream.
4.3.12.3.29 loudmntry - momentary loudness
This 11-bit element indicates the momentary loudness of the programme without any gain adjustments due to dialogue normalization and dynamic range compression applied, measured since the previous value was sent in the bitstream. The loudmntry measurement is specified in Recommendation ITU-R BS.1771 [i.4] (or EBU Tech 3341 [i.11]).
The loudmntry parameter is encoded according to:
loudmntry = �momentary_loudness×10+0,5� + 1024
and decoded according to: ��������_�������� = (loudmntry − 1024)/10[LUFS] 4.3.12.3.30 b_max_loudmntry - maximum momentary loudness flag
If true, this Boolean indicates that a max_loudmntry field follows in the bitstream.
4.3.12.3.31 max_loudmntry - maximum momentary loudness
This 11-bit element indicates the maximum momentary loudness of the programme measured as specified in Recommendation ITU-R BS.1771 [i.4] (or EBU Tech 3341 [i.11]) without any gain adjustments due to dialnorm and dynamic range compression applied.
The max_loudmntry parameter is encoded according to:
max_loudmntry = �maximum_momentary_loudness×10+0,5� + 1024
and decoded according to: �������_��������_�������� = (max_loudmntry − 1024)/10[LUFS] 4.3.12.3.32 b_extension - extension flag
If true, this Boolean indicates that further loudness extension data is present.
4.3.12.3.33 e_bits_size - extension size
This 5-bit element, potentially extended with variable_bits(), indicates the size of the extension_bits field in bits.
4.3.12.3.34 extension_bits - extension bits
This field of size e_bits_size holds extension bits. The content of these bits is not defined in the present document.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)110
4.3.12.4 extended_metadata
4.3.12.4.1 b_associated - associate substream flag - parameter
The setting of this Boolean depends on the following bitstream elements:
• presentation_config
• content_classifier
The b_associated flag shall be true if:
• content_classifier ∈ �0b101; 0b011;0b010�, or
• presentation_config ∈ �2; 4� For all other cases, b_associated shall be false.
4.3.12.4.2 b_dialog - dialogue substream flag - parameter
The setting of this Boolean depends on the following bitstream elements:
• presentation_config
• content_classifier
The b_dialog flag shall be TRUE for several specific cases that are defined in table 157a. For all other cases, b_dialog shall be false.
Table 157a: b_dialog
presentation_config content_classifier b_dialog Note 0 0bxxx true The given substream has to be the dialogue substream of the
corresponding presentation. 3 0bxxx true The given substream has to be the dialogue substream of the
corresponding presentation. x 0b100 true
NOTE: The 'x' is indicating any valid value is allowed.
4.3.12.4.3 b_scale_main - scale main flag
If true, this Boolean indicates that scale main information is present.
4.3.12.4.4 scale_main - scale main
This 8-bit element represents a negative gain of 0 dB (0x00) to 76,2 dB (0xfe) in 0,3 dB steps. A value of 0xff indicates a full mute.
4.3.12.4.5 b_scale_main_center - scale main centre flag
If true, this Boolean indicates that scale main centre information is present.
4.3.12.4.6 scale_main_center - scale main centre
This 8-bit element represents a negative gain of 0 dB (0x00) to 76,2 dB (0xfe) in 0,3 dB steps. A value of 0xff indicates a full mute.
4.3.12.4.7 b_scale_main_front - scale main front flag
If true, this Boolean indicates that scale main front information is present.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)111
4.3.12.4.8 scale_main_front - scale main front
This 8-bit element represents a negative gain of 0 dB (0x00) to 76,2 dB (0xfe) in 0,3 dB steps. A value of 0xff indicates a full mute.
4.3.12.4.9 pan_associated - associate pan data- 8 bits
This 8-bit value represents an angle from 0 (0x00) to 358,5 (0xef) degrees in 1,5 degree steps. The values 0xf0…0xff shall not be used. The angle is to be interpreted as 0 pointing in the front direction, and increases in the clockwise direction, for example, standard Right speaker is at +30 degree (0x14).
4.3.12.4.10 b_dialog_max_gain - dialog maximum gain flag
If true , this Boolean indicates that a dialog_max_gain element is present and shall follow in the bitstream.
4.3.12.4.11 dialog_max_gain - dialogue maximum gain
This 2-bit element indicates the maximum gain, g_dialog_max, by which a user might change the balance between music and effects (M+E) and dialogue in favour of dialogue. The g_dialog_max value is determined as follows:
g_dialog_ max�dB� = (1 + ������_max_����) × 3
4.3.12.4.12 b_pan_dialog_present - dialogue pan data present flag
If true, this Boolean indicates that dialogue pan data is present and shall follow in the bitstream.
4.3.12.4.13 pan_dialog - dialogue pan data
This 8-bit element indicates the angle used for panning one or two dialogue signals onto the music and effects (M&E) track. The meaning is the same as for pan_associated. See clause 4.3.12.4.9.
4.3.12.4.14 pan_signal_selector - selector of panning signal
Table 158: pan_signal_selector
pan_signal_selector value Meaning 0 Full 3ch decode 1 Decode sf_data 1 and 2 2 Decode sf_data 0 and 2 3 Decode sf_data 0 and 1
This 2-bit field shall be evaluated only if all of the following conditions are met:
• channel_mode is 2 (indicating three channels);
• 3_0_channel_element is used;
• 3_0_codec_mode is set to SIMPLE; and
• 3_0_coding_config is set to 1.
If all of the above conditions are met and the value of pan_signal_selector is larger than 0, then full decoding of the three-channel element and processing (as described in clause 5.3.3.3) can be simplified with the following operations:
• partial decoding of three_channel_data as indicated in table 158;
• omission of the processing described in clause 5.3.3.3; and
• panning of the two decoded signals using pan_dialog[0] and pan_dialog[1].
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)112
4.3.12.4.15 b_channels_classifier - channel classifier flag
If true, this Boolean indicates that channel classification data is available.
4.3.12.4.16 b_{c,l,r,ls,rs,lrs,rrs,lw,rw,vhl,vhr,lfe}_active - channel active flag
If true, this Boolean indicates that the corresponding channel, C, L, R, Ls, Rs, Lrs, Rrs, Lw, Rw, Vhl, Vhr, or LFE, is active.
4.3.12.4.17 b_{c,l,r}_has_dialog - channel has dialogue flag
If true, this Boolean indicates that the corresponding channel, C, L or R, contains a dialogue.
4.3.12.4.18 b_event_probability_present - event probability present flag
If true, this Boolean that an event probability value is transmitted.
4.3.12.4.19 event_probability - event probability
This 4-bit element indicates the probability that a congruent audio event is ongoing. A high value corresponds to a high certainty.
4.3.12.5 Channel mode query functions
4.3.12.5.1 channel_mode_contains_Lfe()
This function returns true if the channel mode contains an Low-Frequency Effects channel.
Pseudocode 9
channel_mode_contains_Lfe() { if (ch_mode == 4 || ch_mode == 6 || ch_mode == 8 || ch_mode == 10) { return TRUE; } return FALSE; }
4.3.12.5.2 channel_mode_contains_c()
This function returns true if the channel mode contains a Centre channel.
Pseudocode 10
channel_mode_contains_c() { if (ch_mode == 0 || (ch_mode >= 2 && ch_mode <= 10)) { return TRUE; } return FALSE; }
4.3.12.5.3 channel_mode_contains_lr()
This function returns true if the channel mode contains a Left and a Right channel.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)113
Pseudocode 11
channel_mode_contains_lr() { if (ch_mode >= 1 && ch_mode <= 10) { return TRUE; } return FALSE; }
4.3.12.5.4 channel_mode_contains_LsRs()
This function returns true if the channel mode contains a Left Surround and a Right Surround channel.
Pseudocode 12
channel_mode_contains_LsRs() { if (ch_mode >= 3 && ch_mode <= 10) { return TRUE; } return FALSE; }
4.3.12.5.5 channel_mode_contains_LrsRrs()
This function returns true if the channel mode contains a Left Rear Surround and a Right Rear Surround channel.
Pseudocode 13
channel_mode_contains_LrsRrs() { if (ch_mode == 5 || ch_mode == 6) { return TRUE; } return FALSE; }
4.3.12.5.6 channel_mode_contains_LwRw()
This function returns true if the channel mode contains a Left Wide and a Right Wide channel.
Pseudocode 14
channel_mode_contains_LwRw() { if (ch_mode == 7 || ch_mode == 8) { return TRUE; } return FALSE; }
4.3.12.5.7 channel_mode_contains_VhlVhr()
This function returns true if the channel mode contains a Left Vertical Height and a Right Vertical Height channel.
Pseudocode 15
channel_mode_contains_VhlVhr() { if (ch_mode == 9 || ch_mode == 10) { return TRUE; } return FALSE; }
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)114
4.3.13 Dynamic range control - DRC
4.3.13.1 drc_frame - dynamic range control
4.3.13.1.1 b_drc_present - DRC present
If true, this Boolean indicates that DRC data is available in the bitstream.
4.3.13.2 drc_config - DRC configuration
4.3.13.2.1 drc_decoder_nr_modes - number of DRC decoder modes
This 3-bit element indicates the number of DRC decoder modes carried in drc_frame, which is drc_decoder_nr_modes + 1.
4.3.13.2.2 drc_eac3_profile - (E-)AC-3 profile
This 3-bit element indicates which (E-)AC-3 profile to use when transcoding an AC-4 stream into (E-)AC-3. Table 159 shows how the values of this field are related to the (E-)AC-3 profile.
Table 159: (E-)AC-3 profiles
drc_eac3_profile Profile 0 None 1 Film standard 2 Film light 3 Music standard 4 Music light 5 Speech 6 Reserved 7 Reserved
4.3.13.3 drc_decoder_mode_config - DRC decoder mode configuration
4.3.13.3.1 drc_decoder_mode_id - DRC decoder mode ID
This 3-bit element indicates DRC decoder mode IDs as shown in table 160.
Table 160: drc_decoder_mode_id
Value of drc_decoder_mode_id DRC decoder mode Output level range 0 Home Theatre -31...-27 1 Flat panel TV -26...-17 2 Portable - Speakers -16...0 3 Portable - Headphones -16...0 4 Reserved - 5 Reserved - 6 Reserved - 7 Reserved -
4.3.13.3.2 drc_output_level_from - lowest reference output level
This 5-bit element indicates the lowest output level of the reference level (dialnorm, see clause 4.3.12.2.1) for the corresponding DRC decoder mode. The value of the field represents levels in dBFS according to:
Lout,min = −drc_output_level_from [dB]
The output range for the default DRC decoder modes is defined in table 160.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)115
4.3.13.3.3 drc_output_level_to - highest reference output level
This 5-bit element indicates the highest output level of the reference level (dialnorm, see clause 4.3.12.2.1) for the corresponding DRC decoder mode.
The value of the field represents levels in dBFS according to:
Lout,max = −drc_output_level_to [dB]
The output range for the default DRC decoder modes is defined in table 160.
4.3.13.3.4 drc_repeat_profile_flag - repeat profile flag
This bit indicates the presence of the drc_repeat_id field as described in clause 4.3.13.3.5.
4.3.13.3.5 drc_repeat_id - repeat data from ID
When the configuration of two or more DRC decoder modes is identical, this 3-bit element provides an option to duplicate the configuration without signalling it twice.
If this field is present, this DRC decoder mode is defined by the DRC decoder mode with ID drc_repeat_id.
4.3.13.3.6 drc_default_profile_flag - default profile flag
This bit indicates that the (E-)AC-3 profile [i.1], indicated by drc_eac3_profile, shall be used to define the compression curve and time constants. The definition of the (E-)AC-3 profiles data is given in table 161.
NOTE 1: Semantics for the compression curve parameters are given in clause 4.3.13.4.
Table 161: (E-)AC-3 profiles definition - compression curve parameters
Parameter Profile None Film standard Film light Music standard Music light Speech
Compression curve L0,low [dB] 0 0 -10 0 -10 0 L0,high [dB] 0 5 10 5 10 5 Gmaxboost [dB] 0 6 6 12 12 15 drc_nr_boost_sections 0 0 0 0 0 0 Lmaxboost [dB] 0 -12 -22 -24 -34 -19 Gmaxcut [dB] 0 -24 -24 -24 -15 -24 drc_nr_cut_sections 0 1 1 1 0 1 Lsectioncut [dB] 0 15 20 15 - 15 Lmaxcut [dB] 0 35 40 35 40 35 Gsectioncut [dB] 0 -5 -5 -5 - -5 Time smoothing τattack [ms] N/A 100 100 100 100 100 τrelease [ms] N/A 3 000 3 000 10 000 3 000 1 000 τattack,fast [ms] N/A 10 10 10 10 10 τrelease,fast [ms] N/A 1 000 1 000 1 000 1 000 200 drc_attack_threshold N/A 15 15 15 15 10 drc_release_threshold N/A 20 20 20 20 10
NOTE 2: Profile None has a constant gain (i.e. it effectively disables the compressor). Therefore, the time constants are not applicable. Time constants for this profile can be signalled using drc_tc_default_flag = 1.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)116
Figure 1: Parameterization of gain curve
Figure 2: Music and speech compressor profiles
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)117
Figure 3: Film compressor profiles
4.3.13.3.7 drc_compression_curve_flag - compression curve flag
This bit indicates whether a compression curve is transmitted for the current DRC decoder mode (drc_compression_curve_flag = 1) or whether compression gains are transmitted by drc_frame (drc_compression_curve_flag = 0).
4.3.13.3.8 drc_gains_config - DRC gains configuration
This 2-bit element indicates the manner in which configuration DRC gains for this DRC decoder mode are transmitted, and implicitly the total number of DRC bands used in this mode. The number of parameter bands can be derived using table 162, and the definitions for configurations 2 and 3 are listed in table 163.
Table 162: Description of drc_gains_config
Value of drc_gains_config Gains configuration nr_drc_bands 0 Single wideband gain for all channels 1 1 Channel dependent gains, wideband 1 2 Channel dependent gains, 2 frequency bands 2 3 Channel dependent gains, 4 frequency bands 4 NOTE: Channel dependent DRC gains are transmitted for groups of channels. Clause 4.3.13.7.1 gives an overview
of the groups per speaker configuration.
Table 163: Parameter band definition for DRC gains
Parameter band drc_gains_config = 2 drc_gains_config = 3 QMF start band QMF end band QMF start band QMF end band
0 0 4 0 0 1 5 Qmax - 1 1 4 2 N/A 5 16 3 N/A 17 Qmax - 1
Qmax depends on the sampling frequency as shown in table 164.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)118
Table 164: Qmax value
Sampling frequency Qmax 44,1 kHz, 48 kHz 64 96 kHz 128 192 kHz 256
4.3.13.4 drc_compression_curve - compression curve parameters
4.3.13.4.1 DRC compression curve parameterization
Figure 1 shows the parameterization of the gain curve. The gain curve is described by control points, defining sections. The control points are described by pairs (level, gain) which are given in dB2. The level is relative to the reference level dialogue normalization, i.e. a level of zero is at the reference level.
The curve has boost sections, where the gain is all positive (or zero), and a cut section, where it is all negative (or zero).
NOTE: Most of the control points can be implicitly defined, as per table 165.
Table 165: Control points
- Transmitted if Value if transmitted Value if not transmitted Gmaxboost Always drc_gain_max_boost N/A Lmaxboost Gmaxboost > 0 Lsectionboost-(1+drc_lev_max_boost) L0,low Gsectionboost drc_nr_boost_sections=1 (1+drc_gain_section_boost) - Lsectionboost drc_nr_boost_sections=1 L0,low-(1+drc_lev_section_boost) L0,low L0,low Always 0-drc_lev_nullband_low N/A L0,high Always 0+drc_lev_nullband_high N/A Gmaxcut Always 0-drc_gain_max_cut N/A Lmaxcut Gmaxcut < 0 Lsectioncut+(1+drc_lev_max_cut) L0,high Gsectioncut drc_nr_cut_sections=1 0-(1+drc_gain_section_cut) - Lsectioncut drc_nr_cut_sections=1 L0,high+(1+drc_lev_section_cut) L0,high
4.3.13.4.2 drc_lev_nullband_low - null band lower boundary
This 4-bit element defines the lower limit of the null-band. See table 165.
4.3.13.4.3 drc_lev_nullband_high - null band higher boundary
This 4-bit element defines the upper limit of the null-band. See table 165.
4.3.13.4.4 drc_gain_max_boost - maximum boost
This 4-bit element defines the maximum boost gain. See table 165.
4.3.13.4.5 drc_lev_max_boost - start of maximum boosting
This 5-bit element defines the upper limit of the maximum boost section. See table 165.
4.3.13.4.6 drc_nr_boost_sections - number of boost sections
This bit controls the number of boost sections. See table 165.
4.3.13.4.7 drc_gain_section_boost - extra boost section gain
This 4-bit element defines an extra control point gain.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)119
4.3.13.4.8 drc_lev_section_boost - extra boost section control point level
This 5-bit element defines an extra control point level.
4.3.13.4.9 drc_gain_max_cut - maximum cut
This 5-bit element defines the maximum cut gain. See table 165.
4.3.13.4.10 drc_lev_max_cut - start of maximum cutting
This 6-bit element defines the upper limit of the maximum cut section. See table 165.
4.3.13.4.11 drc_nr_cut_sections - number of cut sections
This bit controls the number of cut sections. See table 165.
4.3.13.4.12 drc_gain_section_cut - extra cut section gain
This 5-bit element defines an extra control point gain.
4.3.13.4.13 drc_lev_section_cut - extra cut section control point level
This 5-bit element defines an extra control point level.
4.3.13.4.14 drc_tc_default_flag - DRC default time constants flag
This bit indicates whether the default gain smoothing parameters should be used or if different time constants should be transmitted in the bitstream. If drc_tc_default_flag is 1, the smoothing data in table 166 is used.
Table 166: Default smoothing data
Field Value τattack 100 ms τrelease 3 000 ms τattack,fast 10 ms τrelease,fast 1 000 ms drc_adaptive_smoothing_flag 0
4.3.13.4.15 drc_tc_attack - time constant for attacks
This 8-bit element defines the time constant for smoothing during attacks. The time constant is given by:
τattack = drc_tc_attack × 5[ms]
4.3.13.4.16 drc_tc_release - time constant for release
This 8-bit element defines the time constant for smoothing during signal release. The time constant is given by:
τrelease = drc_tc_release × 40[��] 4.3.13.4.17 drc_tc_attack_fast - time constant fast attacks
This 8-bit element defines the time constant for smoothing during fast attacks. The time constant is given by:
τattack,fast = drc_tc_attack_fast × 5[ms]
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)120
4.3.13.4.18 drc_tc_release_fast - time constant for fast release
This 8-bit element defines the time constant for smoothing during fast signal release. The time constant is given by:
τrelease,fast = drc_tc_release_fast × 20[��] 4.3.13.4.19 drc_adaptive_smoothing_flag - adaptive smoothing flag
This bit indicates whether adaptive smoothing is supported by the profile.
4.3.13.4.20 drc_attack_threshold - fast attack threshold
This 5-bit element indicates the threshold in dB for deciding when to switch to the fast attack time constant.
4.3.13.4.21 drc_release_threshold - fast release threshold
This 5-bit element indicates the threshold in dB for deciding when to switch to the fast release time constant.
4.3.13.5 drc_data - DRC frame-based data
4.3.13.5.1 drc_gainset_size_value - gain set data size
This element codes drc_gainset_size. It is a 6-bit element that can be extended by variable_bits(). The value of drc_gainset_size is the size in bits of the following drc_gains element. A decoder may use this information to skip profile data that are not applicable.
4.3.13.5.2 b_more_bits - more bits flag
If true, this Boolean the presence of additional variable_bits to be used for the drc_gainset_size determination.
4.3.13.5.3 drc_version - DRC version
This 2-bit element indicates the version of the DRC gainset data. For DRC gainset data that conforms to the present document, the drc_version field shall be set to 0b00, and thus the drc2_bits field shall not be present in the payload. A decoder implemented in accordance with the present document shall skip the drc2_bits field if the drc_version field is not set to 0b00.
4.3.13.5.4 drc2_bits - DRC gainset extension bits
This element of size bits_left, which is only present if drc_version is not 0, holds additional DRC gainset data. The content of this additional DRC gainset data is not specified here since bitstreams conforming to the present document shall have drc_version=0.
4.3.13.5.5 drc_reset_flag - DRC reset flag
This bit signals a reset of the DRC algorithm.
4.3.13.5.6 drc_reserved - reserved bits
This 2-bit element is reserved for future use and the content shall be skipped by an AC-4 decoder conforming to the present document.
4.3.13.6 drc_gains - DRC gains
4.3.13.6.1 drc_gain_val - DRC gain
This 7-bit element contains the first DRC gain.
drc_gain[0][0][0]=�drc_gain_val − 64� [dB�]
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)121
4.3.13.6.2 drc_gain_code - DRC gain codeword
This variable-length element contains a Huffman code. The huff_decode_diff() function looks up the Huffman code using the Huffman code book DRC_HCB specified in table A.62 and generates a differential DRC gain value.
4.3.13.7 DRC helper elements
4.3.13.7.1 nr_drc_channels - number of DRC channels
The nr_drc_channels variable indicates for how many channels DRC gains are transmitted. The value of this variable depends on the channel configuration for which the content is transmitted, as defined in table 167.
Table 167: nr_drc_channels
Channel configuration nr_drc_channels Group 1 Group 2 Group 3 Mono 1 C - - Stereo 1 L, R - - 5.1 3 L, R, LFE C Ls, Rs 7.1 (3/4/0) 3 L, R, LFE C Ls, Rs, Lrs, Rrs 7.1 (5/2/0) add_ch_base=0 3 L, R, LFE, Lw, Rw C Ls, Rs 7.1 (5/2/0) add_ch_base=1 3 L, R, LFE C Ls, Rs, Lw, Rw 7.1 (3/2/2) add_ch_base=0 3 L, R, LFE, Vhl, Vhr C Ls, Rs 7.1 (3/2/2) add_ch_base=1 3 L, R, LFE C Ls, Rs, Vhl, Vhr
4.3.13.7.2 nr_drc_subframes - number of DRC subframes
The variable nr_drc_subframes indicates the number of DRC subframes within an AC-4 frame. Table 168 lists the supported frame lengths and the corresponding number of DRC subframes.
Table 168: nr_drc_subframes
Frame length Subframe length nr_drc_subframes 384 384 1 512 256 2 768 256 3 960 320 3 1 024 256 4 1 536 256 6 1 920 320 6 2 048 256 8
4.3.14 dialogue enhancement - DE
4.3.14.1 b_de_data_present - dialogue enhancement data present flag
If true, this Boolean indicates that dialogue enhancement data is present in the bitstream. If an audio signal contains dialogue, it is recommended that dialogue enhancement data is present, i.e. b_de_data_present should be true for all frames.
4.3.14.2 de_config_flag - dialogue enhancement configuration flag
This bit indicates whether a frame that is not an independently decodable frame (I-frame) contains dialogue enhancement configuration data.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)122
4.3.14.3 de_config - dialogue enhancement configuration
4.3.14.3.1 de_method - dialogue enhancement method
This 2-bit element indicates the dialog enhancement method as shown in table 169.
Table 169: Description of de_method values
Value de_method 0 Channel independent dialogue enhancement 1 Cross-channel dialogue enhancement 2 Waveform-parametric hybrid; channel independent enhancement and waveform channel 3 Waveform-parametric hybrid; cross-channel enhancement and waveform channel
4.3.14.3.2 de_max_gain - maximum dialogue enhancement gain
This 2-bit element defines the maximum gain (Gmax) allowed for boosting the dialogue:
Gmax = �de_max_gain + 1� × 3 [dB�]
4.3.14.3.3 de_channel_config - channel configuration
This 3-bit element indicates the channels for which dialogue enhancement parameters are transmitted as specified in table 170, column 2. For mono and stereo content, only a subset of the possible codewords are valid, as marked with an 'X' in column 4 and 5. Column 3 lists values of de_nr_channels derived from de_channel_config. de_nr_channels indicates the total number of channels to be processed by the dialogue enhancement tool.
Table 170: Description of de_channel_config values
de_channel_configvalue Dialogue enhancement parameters present for
de_nr_channels Mono Stereo Multichannel
000 'No parameters' 0 X X X 001 Centre 1 X - X 010 Right 1 - X X 011 Right, Centre 2 - - X 100 Left 1 - X X 101 Left, Centre 2 - - X 110 Left, Right 2 - X X 111 Left, Right, Centre 3 - - X
4.3.14.4 de_data - dialogue enhancement data
4.3.14.4.1 de_keep_pos_flag - keep position flag
This binary flag indicates whether the dialogue panning information of the previous frame should be repeated for the current frame. For I-frames, this flag defaults to zero.
4.3.14.4.2 de_mix_coef1_idx, de_mix_coef2_idx - dialogue panning parameters
These 5-bit elements define the mixing of the dialogue signal onto the first, second and potentially third channel processed by the dialogue enhancement tool. Up to two parameters are decoded using the quantization vector defined in table 171. The last coefficient is always derived considering an energy preserving upmix.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)123
Table 171: Quantization vector for mixing coefficient parameters
de_mix_coefX_idx 0 1 2 3 4 5 6 7 de_mix_coefX 0 6,32 × 10-3 10-2 1,79 × 10-2 3,16 × 10-2 5,65 × 10-2 7,87 × 10-2 0,111 de_mix_coefX_idx 8 9 10 11 12 13 14 15 de_mix_coefX 0,156 0,218 0,303 0,37 0,448 0,533 0,577 0,622 de_mix_coefX_idx 16 17 18 19 20 21 22 23 de_mix_coefX 0,7071 0,783 0,846 0,894 0,929 0,953 0,976 0,9877 de_mix_coefX_idx 24 25 26 27 28 29 30 31 de_mix_coefX 0,9938 0,9969 0,9984 0,9995 0,99984 0,99995 0,99998 1
4.3.14.4.3 de_keep_data_flag - keep data flag
This bit indicates whether the latest transmitted parametric data shall be repeated for the current frame.
4.3.14.4.4 de_ms_proc_flag - M/S processing flag
This bit indicates whether the parameters are related to processing on the Mid/Side representation of the signal.
4.3.14.4.5 de_par_code - parameter code
This variable-length element contains a Huffman code indicating a dialogue enhancement parameter, representing either an absolute parameter quantization index or a differential value.
4.3.14.4.6 de_signal_contribution - contribution of the signal to the enhancement
This 5-bit element indicates the contribution of the coded signal in the hybrid mode. It indicates the ratio of the gain that should be contributed by the coded signal.
αc =de_signal_contribution
31
4.3.14.5 Dialogue enhancement helper elements
4.3.14.5.1 de_nr_bands - number of parameter bands
This constant value indicates the number of dialogue enhancement parameter bands, which is always 8 bands. The QMF band grouping for each dialogue enhancement parameter band is shown in table 172.
Table 172: Dialog enhancement parameter band definition
Dialogue enhancement parameter band First QMF band Last QMF band 0 0 0 1 1 1 2 2 3 3 4 6 4 7 10 5 11 16 6 17 26 7 27 40
4.3.14.5.2 de_par[ ][ ] - dialogue enhancement parameter set
Variable that contains the decoded quantization indices to the dialogue enhancement parameters for de_nr_channels and de_nr_bands.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)124
4.3.14.5.3 de_par_prev[ ][ ] - previous dialogue enhancement parameter set
Variable that contains the decoded quantization indices to the dialogue enhancement parameters for the previous frame. If no parameter was defined in the previous frame(i.e. dialogue enhancement is inactive or the respective channel is not used), it shall be assumed to be 0.
4.3.14.5.4 de_abs_huffman(table_idx, code) - absolute parameter Huffman decoding
This is a function that looks up Huffman codes, resulting in a dialogue enhancement parameter quantization index. If table_idx is zero, the Huffman codebook DE_HCB_ABS_0 from table A.58 is used; otherwise, DE_HCB_ABS_1 from table A.60 is used.
Pseudocode 16
de_abs_huffman(table_idx, code) { if (table_idx == 0) { hcb = DE_HCB_ABS_0; } else { hcb = DE_HCB_ABS_1; } cb_idx = huff_decode_diff(hcb, code); return cb_idx; }
4.3.14.5.5 de_diff_huffman(table_idx, code) - differential parameter Huffman decoding
This is a function that looks up Huffman codes, resulting in a differential dialog enhancement parameter quantization index. If table_idx is zero, the Huffman codebook DE_HCB_DIFF_0 from table A.59 is used; otherwise, DE_HCB_DIFF_1 from table A.61.
Pseudocode 17
de_diff_huffman(table_idx, code) { if (table_idx == 0) { hcb = DE_HCB_DIFF_0; } else { hcb = DE_HCB_DIFF_1; } cb_idx = huff_decode_diff(hcb, code); return cb_idx; }
4.3.15 Extensible metadata delivery format - EMDF
4.3.15.0 Introduction
The Extensible Metadata Delivery Format (EMDF) is an extensible structure for metadata delivery. Each payload is tagged with a specific payload identifier to provide an unambiguous indication of the type of data present in the payload. The order of payloads within the EMDF substream is undefined: payloads can be stored in any order, and a parser shall be able to parse the entire EMDF substream to extract relevant payloads and ignore payloads that are either not relevant or are unsupported.
4.3.15.1 emdf_payloads_substream - EMDF payloads substream
4.3.15.1.1 emdf_payload_id - EMDF payload identification
This element identifies the type of payload that follows in the EMDF substream. It is a 5-bit element that can be extended by variable_bits(). The assignment of emdf_payload_id field values to specific payload types is specified in table 173.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)125
Table 173: Defined EMDF payload types
emdf_payload_id EMDF payload type 0 EMDF substream end NOTE: Further payload IDs in the extended range emdf_payload_id≥0 are registered in AC-4 EMDF datatype
registry [i.13].
The final payload in the EMDF substream shall have an emdf_payload_id value of 0, indicating that no additional payloads follow in the EMDF substream. When the value of the emdf_payload_id is 0, all fields in the emdf_payload_config shall be set to 0, and the value of the emdf_payload_size field shall be set to 0.
4.3.15.1.2 emdf_payload_size - size of EMDF payload
This element indicates the size of the following payload. It is an 8-bit element that can be extended by variable_bits(). The value of the emdf_payload_size element shall be equal to the number of bytes in the following payload, excluding the emdf_payload_size field and all fields in emdf_payload_config.
4.3.15.1.3 emdf_payload_byte - EMDF payload byte
The 8-bit element, together with emdf_payload_size, forms the EMDF payload.
The sequence of emdf_payload_size and emdf_payload_byte values forms the EMDF payload.
4.3.15.2 emdf_payload_config - EMDF payload configuration
4.3.15.2.1 b_smpoffst - payload sample offset flag
If true, this Boolean indicates that the smpoffst field shall follow in the bitstream. If the current payload applies to the first sample of the AC-4 frame, the value of this Boolean shall be false.
4.3.15.2.2 smpoffst - payload sample offset
This element indicates the offset, in units of PCM audio samples, from the beginning of the AC-4 frame to the first PCM audio sample that the data in the payload shall apply to. It is an 11-bit element that can be extended by variable_bits().
NOTE: This field may indicate a sample index that extends beyond the current AC-4 frame.
4.3.15.2.3 b_duration - payload duration flag
If true, this Boolean indicates that the duration field shall follow in the bitstream. If the current payload applies to all samples up to and including the final sample of the AC-4 frame, the value of this Boolean shall be false.
4.3.15.2.4 duration - payload duration - variable_bits(11)
This element indicates the time period in units of PCM audio samples to which the data in the payload applies. It is an 11-bit element that can be extended by variable_bits().
NOTE: This field can indicate a sample index that extends beyond the current AC-4 frame.
4.3.15.2.5 b_groupid - payload group ID flag
If true, this Boolean indicates that the groupid field shall follow in the bitstream.
4.3.15.2.6 groupid - payload group ID
This element provides a mechanism to indicate to a decoder that specific payloads within the EMDF substream are associated with one another and that the payload data are related. It is a 2-bit element that can be extended by variable_bits(). For payloads that are associated with each other, the groupid field for each payload shall be set to the same value.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)126
4.3.15.2.7 b_codecdata - codec specific data flag
If true, this Boolean enables codec-specific data to be included in the payload configuration. For payload configuration data that conforms to the present document, the value of this Boolean shall be false.
4.3.15.2.8 codecdata - codec specific data
This 8-bit element is not specified here since bitstreams that conform to the present document shall have a value of false for b_codecdata.
4.3.15.2.9 b_discard_unknown_payload - discard unknown payload during transcode flag
This Boolean indicates whether the current payload is retained or discarded during a transcoding process if the transcoder does not recognize the specific emdf_payload_id value of the payload.
If true, the current payload shall be discarded during a transcoding process if the emdf_payload_id value is unrecognized.
If false, the current payload shall be retained during a transcoding process unless otherwise indicated by the proc_allowed field.
4.3.15.2.10 b_payload_frame_aligned - payload to audio data frame alignment flag
This Boolean is used to indicate how closely the payload data and the audio data within the AC-4 frame need to be synchronized after decoding and/or during transcoding.
If true, the payload data shall be applied to the decoded audio from that AC-4 frame.
If false, the application of the payload data to the decoded audio from the AC-4 frame is not required to be frame aligned.
4.3.15.2.11 b_create_duplicate - create duplicate payload during transcode flag
In some transcoding operations, the frame length of the output format may be shorter than the length of an AC-4 frame.
This Boolean shall set to true to indicate that the payload shall be duplicated in all frames of the output format that correspond to the same period of time as the current AC-4 frame.
The This Boolean shall be set to false if the payload does not need to be repeated in all frames of the output format that correspond to the same period of time as the current AC-4 frame.
4.3.15.2.12 b_remove_duplicate - remove duplicate payload during transcode flag
This Boolean shall be set to true to indicate that payloads with the same emdf_payload_id value as the current payload need to be included only once in every output frame of the transcoding process when the output frame duration of a transcoding process is longer than the length of an AC-4 frame. Subsequent payloads carried in AC-4 frames that correspond to the same period of time as the current frame of the output format shall be discarded.
This Boolean shall be set to false to indicate that payloads carried in all AC-4 frames that correspond to the same time period as the current output frame of the transcoding process shall be included in this output frame.
4.3.15.2.13 priority - payload priority
This 5-bit element is used to indicate the priority of this payload in relation to other payloads in the EMDF substream. This field may be used to indicate whether a payload can be discarded during a transcoding process to a lower data rate output format that is not able to support all payloads. A priority value of 0 indicates that the payload has the highest priority of all payloads within the EMDF substream. Higher values of the priority field indicate a lower payload priority. Multiple payloads within the EMDF substream may have the same priority field value.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)127
4.3.15.2.14 proc_allowed - processing allowed
This 2-bit element is used to indicate whether a payload should be retained or discarded during transcoding if processing is applied to the metadata and/or audio data decoded from the AC-4 frame during transcoding. The value of the proc_allowed field shall be set as shown in table 174.
Table 174: Meaning of proc_allowed values
proc_allowed Meaning 00 Payload may be retained only if no processing and no changes to any metadata in the AC-4 frame
occur during transcoding. 01 Payload may be retained if no processing other than sampling frequency conversion of the PCM audio
is applied during transcoding. 10 Payload may be retained if one or more of the following processes, but no other processing, occur
during transcoding: • sampling frequency is converted; • momentary sound elements are mixed with the decoded PCM from the frame; • informational metadata related to the audio data is modified; • the channel configuration of the audio is changed.
11 Payload may be retained regardless of any processing performed during transcoding.
4.3.15.3 emdf_protection - EMDF protection data
4.3.15.3.1 protection_length_primary - length of protection_bits_primary field
This 2-bit element is used to specify the length of the protection_bits_primary field. The value of the protection_length_primary field shall be set as shown in table 175.
Table 175: Meaning of protection_length_primary values
protection_length_primary Length of protection_bits_primary field 00 Reserved 01 8 bits 10 32 bits 11 128 bits
4.3.15.3.2 protection_length_secondary - length of protection_bits_secondary field - 2 bits
This 2-bit element is used to specify the length of the protection_bits_secondary field. The value of the protection_length_secondary field shall be set as shown in table 176.
Table 176: Meaning of protection_length_secondary values
protection_length_secondary Length of protection_bits_secondary field 00 0 bits 01 8 bits 10 32 bits 11 128 bits
4.3.15.3.3 protection_bits_primary - primary EMDF substream protection data - 8 to 128 bits
This element has a variable lenght of between 8 and 128 bits. It contains protection data that can be used to verify the integrity of the EMDF substream and the payload data within the EMDF substream. Calculation of the value of the protection_bits_primary field is implementation dependent and is not defined in the present document.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)128
4.3.15.3.4 protection_bits_secondary - secondary EMDF substream protection data bits
This variable-length element of between 0 and 128 bits contains additional optional protection data that may be used to check the integrity of the EMDF substream and the payload data within the EMDF substream. Calculation of the value of the protection_bits_secondary field is implementation dependent and is not defined in the present document.
4.3.16 High sampling frequency - HSF
4.3.16.1 max_sfb_hsf_ext[i] - HSF extension for number of transmitted scale factor bands for index i
This element indicates the number of scale factor bands to be added to the number of scale factor bands transmitted in asf_psy_info to calculate the total number of scale factor bands max_sfb_hsf in the HSF extension.
NOTE: Clause 4.3.16.2 specifies how to calculate max_sfb_hsf.
4.3.16.2 get_max_sfb_hsf(g) – get max_sfb_hsf for group g
This helper function returns the max_sfb_hsf value for the window group g.
Pseudocode 18
get_max_sfb_hsf(g) { idx = 0; if (frame_len_base >= 1536 && (b_long_frame == 0) && (transf_length[0] != transf_length[1])) { num_windows_0 = (1 << (3-transf_length[0])); if (g >= window_to_group[num_windows_0]) { idx = 1; } } max_sfb_hsf = max_sfb[idx] + max_sfb_ext_hsf[idx]; return max_sfb_hsf; }
5 Algorithmic details
5.1 audio spectral front end
5.1.1 Introduction
The audio spectral front end tool is one of two spectral frontend tools that are used to create spectral lines of audio tracks from sf_data() elements in an AC-4 frame. These spectral lines are typically routed through the stereo and multichannel processing tool before being routed into the Inverse Modified Discrete Cosine Transform (IMDCT).
The ASF tool operates in the following stages:
• Entropy coding of spectral values
• Quantization, reconstruction, and scaling
• Spectral ungrouping
The three stages are detailed in clauses 5.1.2 to 5.1.5.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)129
Data and control interfaces
The input is retrieved from an sf_data() element in the AC-4 frame, classified to be processed by the ASF tool.
spectral_data Huffman-coded spectral data (collection of Huffman codewords and sign bits).
The output from the ASF tool is a vector of spectral lines corresponding to track tr:
sASF,tr Vector of (scaled) spectral lines of track tr. This vector is named spec_reord in the spectral ungrouping tool.
NOTE: The track index tr is based on the occurrence of the sf_data() element in the frame: The first sf_data() element corresponds to tr = 0, 2nd to tr = 1, etc.
The control data needed to decode the spectral_data is contained in the sf_data() element and in the corresponding sf_info() element.
5.1.2 Entropy coding of spectral values
5.1.2.1 Data and control interfaces
The input to the Huffman decoding tool is the asf_spectral_data() part of an sf_data() element:
spectral_data Huffman coded spectral data (collection of Huffman codewords and sign bits). The length of the spectral data is only known after the Huffman decoding process.
The output from the Huffman decoding tool is a vector of quantized spectral lines:
quant_spec Vector of quantized spectral lines.
The bitstream parameter and helper elements used by the Huffman decoding tool are:
sect_cb[g][s] Huffman codebook number for section s in group g. This is part of the asf_section_data() structure.
CB_DIM[cb] Table indicating the Huffman codebook dimension for the Huffman codebook cb. See table A.14.
UNSIGNED_CB[cb]
Table indicating whether Huffman codebook cb is a signed or unsigned codebook. See table A.15.
The Huffman codebook tables in annex A contain more Huffman codebook specific information used by the Huffman decoder.
5.1.2.2 Decoding process
The quantized spectral lines are stored as Huffman encoded values in the bitstream. To retrieve the quantized spectral lines, Huffman decoding shall be applied.
For each Huffman codeword in the bitstream, the used Huffman codebook number sect_cb[g][s] is known from the section data. All the valid Huffman codebooks for the spectral lines are listed in clause A.2 through clause A.12. Each Huffman codeword asf_qspec_hcw can be decoded into a symbol cb_idx by a lookup in the corresponding Huffman codebook. The symbol is the index cb_idx of the codeword entry in the Huffman codebook. It determines the values of multiple quantized spectral lines. Depending on the used Huffman codebook, 2 or 4 quantized spectral lines are the result. All Huffman codewords are decoded in the order specified by the syntax of asf_spectral_data(), see clause 4.2.8.4. The detailed decoding process for one symbol is presented in the following pseudocode.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)130
Pseudocode 19
hcb = ASF_HCB_<sect_cb[g][s]>; cb_idx = huff_decode(hcb, asf_qspec_hcw); if (CB_DIM[hcb] == 4) { quant_spec_1 = INT(cb_idx/cb_mod3) - cb_off; cb_idx -= (quant_spec_1+cb_off)*cb_mod3; quant_spec_2 = INT(cb_idx/cb_mod2) - cb_off; cb_idx -= (quant_spec_2+cb_off)*cb_mod2; quant_spec_3 = INT(cb_idx/cb_mod) - cb_off; cb_idx -= (quant_spec_3+cb_off)*cb_mod; quant_spec_4 = cb_idx - cb_off; } else // CB_DIM[hcb] == 2 { quant_spec_1 = INT(cb_idx/cb_mod) - cb_off; cb_idx -= (quant_spec_1+cb_off)*cb_mod; quant_spec_2 = cb_idx - cb_off; }
The values cb_mod, cb_mod2, cb_mod3 and cb_off are Huffman codebook specific and listed together with the Huffman codebook tables in annex A. Huffman codewords using an unsigned Huffman codebook are followed by up to CB_DIM[hcb] sign bits in the bitstream. For each quantized spectral line different to 0, a sign bit is transmitted. A value of 1 indicates that the quantized spectral line is negative. No sign bits are transmitted if a signed Huffman codebook is in use.
If the Huffman codebook number hcb is 0 for the spectral line to be decoded, no Huffman data is stored in asf_spectral_data and the quantized spectral lines are set to 0.
If the Huffman codebook 11 is used and the magnitude value of at least one of the quantized spectral lines would decode to a value of 16 then additional bits are stored after the possible sign bits. For each quantized spectral line with a preliminary magnitude of 16 an extension code is used to determine the real magnitude of the corresponding quantized spectral line. The ext_decode() function for decoding the extension code ext_code is given in the pseudocode below.
Pseudocode 20
ext_decode(ext_code) { N_ext = 0; b = get_bits(ext_code, 1); while (b) { N_ext += 1; b = get_bits(ext_code, 1); } ext_val = get_bits(ext_code, N_ext + 4); return 2**(N_ext + 4) + ext_val; } // get_bits(bs_elem, num_bits) returns the next num_bits in bitstream order from // bitstream element bs_elem as an unsigned value
5.1.3 Quantization reconstruction and scaling
5.1.3.1 Data and control interfaces
The input to the quantization reconstruction and scaling tool is a vector of quantized spectral lines and the asf_scalefac_data() part of an sf_data() element:
quant_spec Vector of quantized spectral lines.
NOTE 1: The maximum allowed value of the elements of quant_spec is 8 191.
scale_factor_data Huffman-coded scale factor difference data (collection of Huffman codewords).
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)131
The output from the quantization reconstruction and scaling tool is a vector of scaled spectral lines:
scaled_spec Vector of scaled spectral lines.
The bitstream parameter and helper elements used by the quantization reconstruction and scaling tool are:
reference_scale_factor reference_scale_factor value as stored in asf_scalefac_data.
sfb_cb[g][sfb] Huffman codebook number for scale factor band sfb in group g.
NOTE 2: This is a helper element derived from the asf_section_data().
max_quant_idx[g][sfb] Maximum of the absolute values of the quantized spectral lines for group g and scale factor band sfb.
NOTE 3: This is a helper element derived during the parsing of asf_spectral_data() and, if b_hsf_ext is true, the spectral data part of ac4_hsf_ext_substream().
5.1.3.2 Decoding process
To convert the quantized spectral lines into scaled spectral lines, a quantization reconstruction and scaling step shall be applied.
The following (non-uniform) reconstruction process shall be used to apply the inverse operation to the non-uniform quantization operation used by the encoder for the spectral lines. The reconstruction stage operates on the output of the Huffman decoder (i.e. on the quantized spectral lines). A reconstructed spectral line rec_spec is calculated from the quantized spectral line quant_spec according to the following formula:
rec_spec=sign�quant_spec� × |quant_spec|��
This reconstruction process shall be done for all spectral lines.
The coded spectrum is divided into scale factor bands. Each scale factor band contains spectral lines which are scaled using a common scale factor gain sf_gain[g][sfb]. A scaled spectral line scaled_spec, which belongs to scale factor band sfb in group g, is calculated from the reconstructed spectral coefficient rec_spec according to the formula:
scaled_spec=sf_gain[g][sfb]×rec_spec
No scale factor is transmitted for scale factor bands that have only zero-valued spectral lines as either indicated by the use of the Huffman codebook 0 for the scale factor band or by the fact that all quantized spectral lines within the scale factor band have an absolute value of zero. The transmitted scale factor values are Huffman coded values of the difference to the previously transmitted scale factor value or to the reference_scale_factor for the first transmitted scale factor difference. The used Huffman codewords for the scale factor difference values are given in table A.1. To get the value of the nth transmitted scale factor sfn, the following formula shall be used:
sfn = sf(���) + cw_idxn − 60
where:
• cw_idxn is the corresponding index in table A.1 to the nth codeword from the bitstream;
• sf(n-1) is the previous scale factor value (or the reference_scale_factor in case n = 1);
• 60 is the index offset to be used in connection with the scale factor codebook.
NOTE: Only scale factor values sfn in the range 0 to 255 are valid scale factor values.
To get a scale factor gain ksf from a scale factor value sf, the following formula shall be used:
ksf = 2�,��∙(����)
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)132
All decoded scale factor gain values ksf are stored in sf_gain[g][sfb] during the decoding of the asf_scalefac_data() element. This is shown in the following pseudocode, which takes the bitstream grouping into account.
Pseudocode 21
scale_factor = reference_scale_factor; first_scf_found = 0; for (g = 0; g < num_window_groups; g++) { max_sfb_g = get_max_sfb(g); for (sfb = 0; sfb < max_sfb_g; sfb++) { if (sfb_cb[g][sfb] != 0) { if (max_quant_idx[g][sfb] > 0) { if (first_scf_found == 1) { dpcm_sf[g][sfb] = huff_decode(ASF_HCB_SCALEFAC, asf_sf_hcw); scale_factor += dpcm_sf[g][sfb] - 60; else { first_scf_found = 1; } sf_gain[g][sfb] = pow (2.0, 0.25 * (scale_factor - 100)); } } } }
5.1.4 Spectral noise fill
5.1.4.1 Data and control interfaces
The input to and the output from the spectral noise fill tool is a vector of spectral lines in bitstream order:
scaled_spec Vector of (scaled) spectral lines in bitstream order.
The bitstream parameter and helper elements used by the spectral noise fill tool are:
num_window_groups Number of window groups.
NOTE 1: This is an sf_info helper element.
num_win_in_group[g] Vector indicating the number of windows in group g.
sfb_offset[sfb] Vector indicating the scale factor band offset for scale factor band sfb.
The matching sfb_offset[sfb] vector for the given sampling frequency and transform length related to the current window shall be used. All possible sfb_offset[sfb] vectors are listed as tables in annex B.
sfb_cb[g][sfb] Huffman codebook number for scale factor band sfb in group g.
NOTE 2: This is a helper element derived from the asf_section_data().
max_quant_idx[g][sfb] Maximum of the absolute values of the quantized spectral lines for group g and scale factor band sfb.
NOTE 3: This is a helper element derived during the parsing of asf_spectral_data() and, if b_hsf_ext is true, the spectral data part of ac4_hsf_ext_substream().
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)133
5.1.4.2 Decoding process
Spectral lines in scaled_spec that are 0 because either the corresponding sfb_cb[g][sfb] is 0 or the corresponding max_quant_idx[g][sfb] is 0 shall be replaced by random noise of a transmitted level if noise fill data is present as indicated when b_snf_data_exists is false. If b_snf_data_exists is false, the spectral noise fill tool is not active and scaled_spec is not modified.
The first step in decoding the spectral noise fill data is to find the initial noise fill reference level. This is the first decoded band that has at least one non-zero component. The following pseudocode shows this process.
Pseudocode 22
k = 0; previous_rms = -1000; /* init to a suitably large negative value */ for (g = 0; g < num_window_groups; g++) { max_sfb_g = get_max_sfb(g); for (sfb = 0; sfb < max_sfb_g; sfb++) { band_rms = 0; for (w = 0; w < num_win_in_group[g]; w++) { for (l = sfb_offset[sfb]; l < sfb_offset[sfb+1]; l++) { band_rms += scaled_spec[k]* scaled_spec[k]; k++; } } if (band_rms > 0) { band_rms /= num_win_in_group[g] * (sfb_offset[sfb+1] - sfb_offset[sfb]); previous_rms = 1.44269504 * log(band_rms); break; } } if (previous_rms != -1000) break; }
Next, the noise fill delta Huffman codewords are read from the bitstream, decoded and the appropriate noise fill levels are computed. This decoding process and the insertion of the noise signal are shown in the following pseudocode.
Pseudocode 23
k = 0; for (g = 0; g < num_window_groups; g++) { max_sfb_g = get_max_sfb(g); for (sfb = 0; sfb < max_sfb_g; sfb++) { band_rms = 0; for (w = 0; w < num_win_in_group[g]; w++) { for (l = sfb_offset[sfb]; l < sfb_offset[sfb+1]; l++) { band_rms += scaled_spec[k]* scaled_spec[k]; k++; } } if (band_rms > 0) { band_rms /= num_win_in_group[g] * (sfb_offset[sfb+1] - sfb_offset[sfb]); previous_rms = 1.44269504 * log(band_rms); } else { /* noise fill band */ dpcm_snf[g][sfb] = huff_decode(ASF_HCB_SNF, asf_snf_hcw); delta = dpcm_snf[g][sfb] - 17; if (delta != -17) { /* -17 is an escape for no noise fill */ /* and the relative level is NOT updated */ noise_rms = previous_rms + delta; previous_rms = noise_rms; noise_rms = pow(2.0, 0.5*noise_rms); k -= num_win_in_group[g] * (sfb_offset[sfb+1] - sfb_offset[sfb]) for (w = 0; w < num_win_in_group[g]; w++) { for (l = sfb_offset[sfb]; l < sfb_offset[sfb+1]; l++) { noise = GetRandomNoiseValue(&nRndStateSnf) * noise_rms; scaled_spec[k++] = noise; } } } }
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)134
} }
The same GetRandomNoiseValue() function as in the speech spectral front end is used, but with an own state, nRndStateSnf. This function returns a normal distributed random number with unit variance and zero mean. The random number generator is initialized at the beginning of the decoding of an ASF frame, using the sequence_counter value, by calling the function ResetRandGenStateSnf(&nRndStateSnf, sequence_counter). This function is described in the following pseudocode.
Pseudocode 24
// initialize / reset the random number generator for spectral noise fill void ResetRandGenStateSnf(ssf_rndgen_state *psS, int sequence_counter) { int stateIdx; int currentIdx; int start = 255 * (sequence_counter % 256); psS->uiOffsetA = start % 255; psS->uiOffsetB = (start / 255) % 256; stateIdx = (psS->uiOffsetA+1)*((psS->uiOffsetA+2)+2*psS->uiOffsetB)/2 + 255*psS->uiOffsetB *(255+psS->uiOffsetB)/2; if (start % 130560 >= 65280) stateIdx += 128; psS->uiStateIdx = stateIdx % 256; currentIdx = psS->uiOffsetA*(psS->uiOffsetA+1)/2 + psS->uiOffsetB*32386; psS->uiCurrentIdx = currentIdx % 256; }
5.1.5 Spectral ungrouping tool
5.1.5.1 Data and control interfaces
The input to the ungrouping tool is a vector of spectral lines in bitstream order:
scaled_spec Vector of (scaled) spectral lines in bitstream order.
The output from the ungrouping tool is a vector of reordered spectral lines:
spec_reord Vector of reordered (scaled) spectral lines, i.e. ordered according to the window number and ascending within a window.
The bitstream parameters used by the ungrouping tool are:
num_window_groups Number of window groups.
NOTE: This is an sf_info helper element.
num_win_in_group[g] Vector indicating the number of windows in group g.
sfb_offset[sfb] Vector indicating the scale factor band offset for scale factor band sfb.
The matching sfb_offset[sfb] vector for the given sampling frequency and transform length related to the current window shall be used. All possible sfb_offset[sfb] vectors are listed as tables in annex B.
win_offset[win] Vector indicating the offset to the start of window win in the reordered output.
5.1.5.2 Decoding process
When more than one transform block is used in an AC-4 frame, spectral lines are stored in groups. To undo this grouping, the AC-4 decoder shall apply reordering as specified in the following pseudocode.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)135
Pseudocode 25
k = 0; win = 0; spec_reord[] = 0; /* init all spectral lines with 0 */ for (g = 0; g < num_window_groups; g++) { max_sfb = get_max_sfb(g); for (sfb = 0; sfb < max_sfb; sfb++) { for (w = 0; w < num_win_in_group[g]; w++) { for (l = sfb_offset[sfb]; l < sfb_offset[sfb+1]; l++) { spec_reord[win_offset[win+w] + l] = scaled_spec[k++]; } } } win += num_win_in_group[g]; }
5.2 Speech spectral frontend (SSF)
5.2.1 Introduction
The speech spectral frontend is an alternative spectral frontend to the audio spectral front end. It is designed to be used for speech content. Like the ASF, the SSF provides blocks of spectral lines that need to be inverse transformed using the same inverse MDCT as for the ASF output. The SSF data that is part of an AC-4 frame is subdivided into one or two parts, called SSF granules.
Each SSF granule is either independently decodable and called an SSF-I-frame or the SSF granule relies on decoded data from previous SSF granules and is called an SSF-P-frame. The decoding of SSF data can only start with an independently decodable frame. There are two possible operation modes of the SSF decoder: the so called SHORT_STRIDE mode, where the SSF granule is subdivided into 4 blocks and the so called LONG_STRIDE mode in which the SSF granule contains just one block. One feature of the SHORT_STRIDE mode is that it includes an envelope interpolation.
Data and control interfaces
The input is retrieved from an sf_data() element in the AC-4 frame, classified to be processed by the SSF tool.
speech_data Envelope and gain indices encoded in the bitstream, as well as decoded envelope values related to the previous frame in the case of SSF-P-frames.
The output from the SSF tool is a vector of spectral lines:
sSSF,ch Vector of n_mdct (scaled) spectral lines of track ch.
NOTE: The track index ch is based on the occurrence of the sf_data() element in the frame: the first sf_data() element corresponds to ch = 0, the second to ch = 1, etc.
The control data needed to decode the speech_data is contained in the sf_data() element.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)136
5.2.2 Top level structure of the SSF
Figure 4: SSF decoder block diagram
Figure 4 shows the SSF decoder block diagram. The predictor decoder, the spectrum decoder and the envelope decoder take bitstream elements as input. The output of the spectrum decoder is combined (added) with the output of the subband predictor and sent to the inverse flattening which applies the signal envelope from the envelope decoder. The signal after the inverse flattening, filled with additional zero spectral lines to provide n_mdct spectral lines, is the output signal of the SSF. This signal is also the input of the subband predictor which uses buffered SSF output signals, the predictor parameters from the predictor decoder and the signal envelope from the envelope decoder to generate the subband predictor output. The envelope allocation and the predictor parameters are additional inputs of the spectrum decoder.
Each block of the SSF decoder block diagram is explained in more detail in the following clauses.
5.2.3 Envelope decoder
5.2.3.0 Data and control interfaces
The input to the envelope decoder is a vector of envelope indices, a vector of gain indices and a vector decoded envelope values from the previous frame:
env_idx[band] Vector of num_bands envelope indices.
NOTE 1: The first index describes an independently coded value of the first band while the remaining indices describe differentially coded envelope values.
gain_idx[block] Vector of num_blocks gain indices.
env_prev[band]
Vector of num_bands decoded envelope values related to the previous frame.
NOTE 2: For SSF-P-frames, this envelope is the env[band] envelope from the previous frame. For SSF-I-frames, this is the decoded envelope indices related to the extra startup envelope that is transmitted in the SHORT_STRIDE case.
The outputs of the envelope decoder are two arrays:
env_alloc[block][band] num_blocks × num_bands dimensional array of envelope allocation values (integer).
f_env_signal[block][band]
num_blocks × num_bands dimensional array of signal envelope values (float).
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)137
The bitstream parameters used by the envelope decoder are:
num_bands Number of bands.
num_blocks Number of blocks. For the LONG_STRIDE mode, num_blocks = 1, and for the SHORT_STRIDE mode, num_blocks = 4.
The following intermediate signals are used by the envelope decoder:
env[band] Vector of num_bands decoded envelope values.
env_interp[block][band]
num_blocks × num_bands dimensional array of interpolated envelope values. gain[block]
Vector of num_blocks decoded gain values.
5.2.3.0a Envelope decoding
Pseudocode 25a: Decoding of env_idx[band] into env[band]
// decode envelope ENV_DELTA_MIN = -16; ENV_BAND0_MIN = -28; env[0] = env_idx[0] + ENV_BAND0_MIN; for (band = 1; band < num_bands; band++) { delta = env_idx[band] + ENV_DELTA_MIN; env[band] = env[band-1] + delta; }
NOTE: Valid env[band] entries are in the range -64, …, 63.
5.2.3.0b Envelope interpolation
For the SHORT_STRIDE mode, envelope interpolation shall be done. For the LONG_STRIDE mode, no interpolation is needed and env_interp[0][band] = env[band]. The envelope interpolation is described in the following pseudocode.
Pseudocode 25b
// interpolate envelope (SHORT_STRIDE) const int32 iUnit = 1024; /* Q.10 */ const int32 iHalf = 512; /* Q.10 */ const int32 iInvNumBlocks = 256; /* Q.10 */ iNumLeftBlocks = SSF_I16_MUL_I16(4U, iUnit); /* Q3.0*Q0.10=Q3.10 */ for (band = 0; band < num_bands; band++) { iLeftDelta = SSF_I16_MUL_I16((env[band] - env_prev[band]), iUnit); /* Q7.10 */ iLeftSlope = SSF_I16_MUL_I16(iLeftDelta, iInvNumBlocks); /* Q7.10*Q0.10=Q7.20 */ iLeftSlope = SSF_I32_SHIFT_RIGHT(iLeftSlope, 10); /* Q7.10 */ for (block = 0; block < num_blocks; block++) { iInterpEnv = SSF_I16_MUL_I16((1 + block), iLeftSlope); /* Q3.0*Q7.10=Q10.10 */ iTmp = SSF_I16_MUL_I16(env_prev[band], iUnit)); iInterpEnv = SSF_I32_ADD_I32(iInterpEnv, iTmp); /* Q10.10 */ if (iInterpEnv > 0) { iInterpEnv = SSF_I32_ADD_I32(iInterpEnv, iHalf); } else { iInterpEnv = SSF_I32_SUB_I32(iInterpEnv, iHalf); } env_interp[block][band] = SSF_I32_SHIFT_RIGHT(iInterpEnv, 10); /* Q.0 */ } }
NOTE: The fixed point operators used are defined in clause 5.2.8.2.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)138
5.2.3.0c Gain decoding
For the SHORT_STRIDE mode, gain indices are transmitted for each of the four blocks. The following pseudocode shows how the gain indices are converted into gains.
Pseudocode 25c: Decoding of gain_idx[block] into gain[block]
// gain_idx to gain conversion for (block = 0; block < 4; block++) { gain[block] = pow(10.0f, gain_idx[block] * 0.1f); }
For LONG_STRIDE, gain_idx[0] = 0 and gain[0] = 1.
5.2.3.0d Envelope refinement
Pseudocode 25d
SSF_HIGH_FREQ_GAIN_THRESHOLD = 2; ENV_MIN = -64; ENV_MAX = 63; // signal envelope for (block = 0; block < num_blocks; block++) { for (band = 0; band < num_bands; band++) { // envelope reconstruction f_env_signal[block][band] = pow(2.0f, 0.5f * env_interp[block][band]); if (band >= SSF_HIGH_FREQ_GAIN_THRESHOLD) { // apply gain f_env_signal[block][band] *= gain[block]; } } } // allocation envelope for (block = 0; block < num_blocks; block++) { for (band = 0; band < num_bands; band++) { env_alloc[block][band] = env_interp[block][band]; if (band >= SSF_HIGH_FREQ_GAIN_THRESHOLD) { // apply gain env_alloc[block][band] += round(2.0f * gain_idx[block] / 3.0f); // limit envelope if (env_alloc[block][band] < ENV_MIN) env_alloc[block][band] = ENV_MIN; if (env_alloc[block][band] > ENV_MAX) env_alloc[block][band] = ENV_MAX; } } }
5.2.4 Predictor decoder
5.2.4.0 Data and control interfaces
The inputs to the predictor decoder are SSF bitstream elements.
The outputs from the predictor decoder are the following predictor parameters:
f_pred_gain[block] Vector of num_blocks predictor gain values.
f_pred_lag[block]
Vector of num_blocks predictor lag values.
The state internal to the predictor decoder is the predictor lag index:
i_prev_pred_lag_idx Predictor lag index of the previous block.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)139
5.2.4.0a Parameter calculation
Pseudocode 25e
// decode predictor parameters for block index block PRED_LAG_DELTA_MIN = -8; if ((block >= start_block) && (block < end_block)) { // predictor parameters are possible if (predictor_presence_flag[block] == 1) { i_pred_gain_idx[block] = arithmetic_decode_pred(); // reconstruct f_pred_gain[block] = PRED_GAIN_QUANT_TAB[i_pred_gain_idx[block]]; if (delta_flag[block] == 1) { i_pred_lag_idx[block] = predictor_lag_delta_bits[block]; i_pred_lag_idx[block] += i_prev_pred_lag_idx + PRED_LAG_DELTA_MIN; } else { i_pred_lag_idx[block] = predictor_lag_bits[block]; } } else { f_pred_gain[block] = 0.0f; i_pred_lag_idx[block] = 0; } } else { //reset predictor parameters f_pred_gain[block] = 0.0f; i_pred_lag_idx[block] = 0; } // update i_prev_pred_lag_idx for next block i_prev_pred_lag_idx = i_pred_lag_idx[block]; // reconstruct f_pred_lag[block] = 640*pow(2,(i_pred_lag_idx[block]-509)/170);
NOTE 1: PRED_GAIN_QUANT_TAB is defined in table C.3.
NOTE 2: Valid i_pred_lag_idx[block] entries are in the range 0,…,509.
5.2.5 Spectrum decoder
5.2.5.1 Spectrum decoder: Data and control interfaces
The inputs to the spectrum decoder are envelope allocation values and the predictor gain for the current block:
env_alloc[band] Vector of num_bands envelope allocation values (integer) for current block.
f_pred_gain Predictor gain value for the current block.
block Block index of the current block.
The output from the spectrum decoder is a vector of residual spectral lines:
f_spec_res[bin] Vector of num_bins residual spectral lines.
The bitstream parameters used by the spectrum decoder are:
num_bands Number of bands.
num_bins Number of coded spectral lines.
n_mdct SSF block length.
alloc_offset_bits Value of bitstream element alloc_offset_bits for the current block.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)140
5.2.5.2 Spectrum decoding process
5.2.5.2.0 Introduction
The process of spectrum decoding comprises the following steps:
1) Helper variable calculation:
The helper variables f_rfu, i_alloc_dithering_threshold, f_adaptive_noise_gain and f_adaptive_noise_gain_var_pres are calculated for the current block. See the pseudocode example in clause 5.2.5.2.1.
2) Heuristic scaling and envelope allocation:
- If f_rfu = 0, env_alloc_mod[band] is set to env_alloc[band] and f_gain_q[band] is set to 1.0.
- If f_rfu is >0 and variance_preserving_flag is 1, env_alloc_mod[band] is set to env_alloc[band] and f_gain_q[band] is set to 1.0.
- If f_rfu is >0 and variance_preserving_flag is 0, heuristic scaling is calculated and the envelope allocation is modified as demonstrated by pseudocode examples in clause 5.2.5.2.2.
3) Lossless decoding. See the pseudocode example in clause 5.2.5.2.3.
4) Inverse quantisation. See the pseudocode example in clause 5.2.5.2.4.
5) If variance_preserving_flag is 0, inverse heuristic scaling is applied. See the pseudocode example in clause 5.2.5.2.5.
5.2.5.2.1 Helper variable calculation
This pseudocode demonstrates how helper variables f_rfu, i_alloc_dithering_threshold, f_adaptive_noise_gain and f_adaptive_noise_gain_var_pres are calculated for the current block.
Pseudocode 26
// compute helper variables f_rfu, i_alloc_dithering_threshold, f_adaptive_noise_gain // and f_adaptive_noise_gain_var_pres for current block RFU_THRESHOLD = 0.75f; ALLOC_DITHERING_THRESHOLD_SMALL = 3; ALLOC_DITHERING_THRESHOLD_LARGE = 5; f_gain = f_pred_gain; if (f_gain < -1.0f) f_rfu = 1.0f; else if (f_gain < 0.0f) f_rfu = -f_gain; else if (f_gain < 1.0f) f_rfu = f_gain; else if (f_gain < 2.0f) f_rfu = 2.0f - f_gain; else // f_gain >= 2.0f f_rfu = 0.0f; if (f_rfu > RFU_THRESHOLD) i_alloc_dithering_threshold = ALLOC_DITHERING_THRESHOLD_SMALL; else i_alloc_dithering_threshold = ALLOC_DITHERING_THRESHOLD_LARGE; if (variance_preserving_flag == 1) i_alloc_dithering_threshold = ALLOC_DITHERING_THRESHOLD_LARGE; f_adaptive_noise_gain_var_pres = sqrt(1 - (f_rfu * f_rfu)); f_adaptive_noise_gain = 1 - f_rfu;
5.2.5.2.2 Heuristic scaling and envelope allocation modification
The pseudocode examples in this clause demonstrate the calculaton of heuristic scaling and envelope allocation modification when f_rfu is >0 and variance_preserving_flag is 0.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)141
Pseudocode 27
// Heuristic scaling and envelope allocation modification ENV_MIN = -64; ENV_MAX = 63; for (band = 0; band < num_bands; band++) { env_in[band] = 3 * env_alloc[band]; } /* compute heuristic scaling */ HeuristicScaling(iRfu, // input: rfu parameter in Qx.10 env_in, // input: 1dB envelope in Qx.0 int_weights_dB); // output in Qx.10 for (band = 0; band < num_bands; band++) {
i_w_dB[band] = SSF_I32_SHIFT_RIGHT(int_weights_dB[band] / 2, 10); // convert result to Q.0
} /* LF-boost */ const int lf_boost_threshold = 3; i_w_dB[0] = (i_w_dB[0] > lf_boost_threshold) ? i_w_dB[0] - lf_boost_threshold : 0; for (band = 0; band < num_bands; band++) { // store gains for Heuristic inverse scaling f_w_dB = FLOAT(int_weights_dB[band]); // conversion from Qx.10 to float f_gain_q[band] = pow(10.0f, 1.5f / 20.0f * f_w_dB); // modify allocation - apply heuristic scaling env_alloc_mod[band] = env_alloc[band] - i_w_dB[band]; // limit allocation envelope to [-64,63] if (env_alloc_mod[band] < ENV_MIN) env_alloc_mod[band] = ENV_MIN; if (env_alloc_mod[band] > ENV_MAX) env_alloc_mod[band] = ENV_MAX; }
The function that performs the heuristic scaling based on a fix point representation of the f_rfu value and the allocation envelope is implemented as below.
Pseudocode 28
// HeuristicScaling() { /* Inputs: iRfu in Qx.10; env_in[] in Q.0 */ /* Output: int_weights_dB[] */ /* constant values in Q.10 */ iDynThreshold = SSF_I32_SHIFT_LEFT(40, 10); iMaxiWdB = SSF_I32_SHIFT_LEFT(15, 10); iInvThree = 341; /* compute max{ env_in } and min{ env_in } */ iMaxEnv = VECMAX(env_in, num_bands);/* find the maximum of the envelope vec. */ iMinEnv = VECMIN(env_in, num_bands);/* find the minimum of the envelope vec. */ iDynUnscaled = iMaxEnv-iMinEnv; iDyn = SSF_I32_SHIFT_LEFT((iMaxEnv-iMinEnv), 10); /* result in Q.10 */ if (iDyn > iDynThreshold) { iCmpFact = iDynThreshold / iDynUnscaled; /* result in Q.10 */ for (band = 0; band < num_bands; band++) { env_local[band] = SSF_I32_SUB_I32(env_in[band], iMinEnv); env_local[band] = SSF_I16_MUL_I16(env_local[band], iCmpFact); } } else { /* make a local copy of the envelope */ for (band = 0; band < num_bands; band++) { env_local[band] = SSF_I32_SUB_I32(env_in[band], iMinEnv); /* Qx.0 */ env_local[band] = SSF_I32_SHIFT_LEFT(env_local[band], 10); /* Q6.10 */ } } /* sort the env_local in descending order */ /* env_local_sorted is a vector with sorted envelope values (Q6.10) */ /* env_indices are such as env_local_sorted = env_local[env_indices] */ Sort(env_local, env_local_sorted, env_indices); /* convert the sorted envelope to a linear domain */ iMtr = 0; for (band = 0; band < num_bands; band++) { /* conversion from dB domain to the log domain is based on a look-up table */ weights_lin[band] = Map_dB_to_Lin(env_local_sorted[band]); /* Q7.10 */ iTmp = SSF_I16_MUL_I16(weights_lin[band],
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)142
band_widths[env_indices[band]); /* Q12.10 */ iMtr = SSF_I32_ADD_I32(iMtr, iTmp); /* Q13.10 + Q13.10 */ } iMtr = SSF_I32_SHIFT_RIGHT(iMtr, 10); iMtr = SSF_I16_MUL_I16(iMtr, iRfu); iMtr = SSF_I32_SHIFT_RIGHT(iMtr, 7); iMtr = SSF_I16_MUL_I16(iMtr, iRfu); iMtr = SSF_I32_SHIFT_RIGHT(iMtr, 3); /* reverse water-filling procedure */ iNum = 0; iMnt = 0; iBsum = 0; while ((iMnt < iMtr) && (band < (num_bands-1))) { iTCurrLev = weights_lin[band]; /* Q7.10 / while ((weights_lin[band] == iTCurrLev) && (band < (num_bands-1))) { iBsum = SSF_I32_ADD_I32(iBsum, band_widths[env_indices[band]); band = band + 1; } iTmp2 = SSF_I32_SUB_I32(iTCurrLev, weights_lin[band]; /* Q7.10 */ iTmp2 = SSF_I16_MUL_I16(iTmp2, iBsum); /* Q7.10 * Q10.0 = Q17.10 */ iMnt = SSF_I32_ADD_I32(iTmp2, iMnt); /* Q17.10 */ } if (iMnt < iMtr) iBsum = num_bins; iTmp = SSF_I32_SUB_I32(iMnt, iMtr); /* Q17.10 */ iTmp = SSF_I32_SHIFT_LEFT(iTmp, 4); iTmp2 = iTmp / iBsum; iTmp2 = SSF_I32_SHIFT_RIGHT(iTmp2, 4); iTCurrLev = SSF_I32_ADD_I32(weights_lin[band], iTmp2); for (band = 0; band < num_bands; band++) { iTmp2 = Map_Lin_to_dB(iTCurrLev); /* Q6.10 */ iTmp = SSF_I32_SUB_I32(env_local[band], iTmp2); /* Q10.10 */ iTmp = SSF_I16_MUL_I16(iTmp, iInvThree); /* Q10.20 */ iTmp = SSF_I32_SHIFT_RIGHT(iTmp, 10); /* Qx.10 */ iTmp = iTmp > 0 ? iTmp : 0; iTmp = iTmp < iMaxiWdB ? iTmp : iMaxiWdB; int_weights_dB[band] = iTmp; /* final result Q10.10 */ } }
The above implementation uses two dedicated functions for an approximate conversion between the linear scale and the dB scale. The conversion from dB values to linear values is implemented by the following function.
Pseudocode 29
int32 Map_dB_to_Lin(int32 iInput // input Qx.10 ) // result in Qx.10 { int32 iRes, iInt, iIndex, iFract; /* this function works with input Q.4, therefore we need a shift */ iInput = SSF_I32_SHIFT_RIGHT(iInput, 6); /* Qx.4 */ /* figure out interval */ iIndex = SSF_I32_SHIFT_RIGHT(iInput, 6); /* Qx.0 */ if (iIndex < 10) { iRes = SSF_I16_MUL_I16(SLOPES_DB_TO_LIN[iIndex], iInput); /* use LUT1 */ /* Q8.4 * max Q11.4 = Q(8+11).8, no overflow by design of input */ iRes = SSF_I32_SHIFT_RIGHT(iRes, 4); iRes = SSF_I32_ADD_I32(iRes, OFFSETS_DB_TO_LIN[iIndex]); /* use LUT2 */ iRes = SSF_I32_SHIFT_LEFT(iRes, 6); /* Qx.10 */ } else { /* index out of range */ iRes = SSF_I32_SHIFT_LEFT(100U, 10); /* Qx.10 */ } return iRes; }
NOTE 1: SLOPES_DB_TO_LIN is specified in table C.13 and OFFSETS_DB_TO_LIN in table C.14.
The conversion from linear values to dB values is implemented by the following function.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)143
Pseudocode 30
int32 Map_Lin_to_dB(int32 iInput // input Qx.10 ) // result in Q7.10 { int32 iRes, iQuantIn, iIndex, iFract, iInt; int32 iTmp1, iTmp2; const int32 iMaxTableSize = 50; iInput = SSF_I32_SHIFT_RIGHT(iInput, 2); // Qx.8 iQuantIn = SSF_I32_SHIFT_RIGHT(iInput, 1); // Qx.8 iIndex = SSF_I32_SHIFT_RIGHT(iQuantIn, 8); // Qx.0 iInt = SSF_I32_SHIFT_LEFT(iIndex, (8 + 1)); iFract = SSF_I32_SUB_I32(iInput, iInt); if (iIndex < iMaxTableSize) { iTmp2 = SSF_I32_SHIFT_LEFT(iIndex, 1); // Qx.0 // access LUT3 iTmp1 = SSF_I16_MUL_I16(SLOPES_LIN_TO_DB[iIndex], iTmp2); // Q11.8 iTmp2 = SSF_I16_MUL_I16(SLOPES_LIN_TO_DB[iIndex], iFract); // Q11.16 iTmp2 = SSF_I32_SHIFT_RIGHT(iTmp2, 8); iRes = SSF_I32_ADD_I32(iTmp1, iTmp2); // access LUT4 iRes = SSF_I32_ADD_I32(iRes, OFFSETS_LIN_TO_DB[iIndex]); // Q11.8 iRes = SSF_I32_SHIFT_LEFT(iRes, 2); } else { iRes = SSF_I32_SHIFT_LEFT(40, 10); // Qx.10 } return iRes; // actually the max here is Q6.10 }
NOTE 2: SLOPES_LIN_TO_DB is specified in table C.15 and OFFSETS_LIN_TO_DB in table C.16.
5.2.5.2.3 Lossless decoding
This pseudocode demonstrates the lossless decoding process.
Pseudocode 31
// Lossless decoding MIN_ALLOC_OFFSET = -21; ENV_MAX_2_MIN_OFFSET = 20; i_alloc_offset = alloc_offset_bits + MIN_ALLOC_OFFSET; // find max entry in env_alloc_mod[band] i_max = env_alloc_mod[0]; for (band = 1; band < num_bands; band++) { if (env_alloc_mod[band] > i_max) i_max = env_alloc_mod[band]; } i_max -= ENV_MAX_2_MIN_OFFSET; // setup i_alloc_table[band] for (band = 0; band < num_bands; band++) { i_alloc_table[band] = env_alloc_mod[band] - i_max + i_alloc_offset; if (i_alloc_table[band] < 0) i_alloc_table[band] = 0; if (i_alloc_table[band] > 20) i_alloc_table[band] = 20; } // setup i_dither[bin] for (band = 0; band < num_bands; band++) { for (bin = start_bin[band]; bin <= end_bin[band]; bin++) { if ((i_alloc_table[band] != 0) && (i_alloc_table[band] < i_alloc_dithering_threshold)) i_dither[bin] = i_dither_cur[block][bin]; else i_dither[bin] = 0; } } arithmetic_decode_coeffs(); // This function uses i_alloc_table[band] and i_dither[bin] // and outputs i_quant_idx[bin]. See clause 5.2.8.2.
5.2.5.2.4 Inverse quantisation
The inverse quantization process is shown in the following pseudocode.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)144
Pseudocode 32
// Inverse quantization (for block index block) /* important variables for bit-exact processing */ int32 i_step_size, i_mid_point, i_dither, i_quant_index; const int32 i_model_unit = (1U << 15); for (band = 0; band < num_bands; band++) { i_alloc = i_alloc_table[band]; /* get the step size */ i_step_size = STEP_SIZES_Q4_15[i_alloc]; for (bin = start_bin[band]; bin <= end_bin[band]; bin++) { if (i_alloc == 0) { f_spec_invq[bin] = GetRandomNoiseValue(&nRndState); if ((variance_preserving_flag[block] == 1) && (band > 1)) { f_spec_invq[bin] *= f_adaptive_noise_gain_var_pres; } else { f_spec_invq[bin] *= f_adaptive_noise_gain; } } else { i_quant_index = i_quant_idx[bin]; if (i_alloc < i_alloc_dithering_threshold) { i_dither = i_dither_cur[block][bin]; i_mid_point = Idx2Reconstruction(i_quant_index, i_dither, i_step_size); f_mid_point = (float)i_mid_point / (float)i_model_unit; f_post_gain = POST_GAIN_LUT[i_alloc-1]; if ((variance_preserving_flag[block] == 1) && (band > 1)) { f_post_gain_var_pres = sqrt(f_post_gain) * f_adaptive_noise_gain_var_pres; if (f_post_gain_var_pres > f_post_gain) { f_post_gain = f_post_gain_var_pres; } } f_spec_invq[bin] = f_mid_point * f_post_gain; } else { // quantizers with no dither i_mid_point = Idx2Reconstruction(i_quant_index, 0, i_step_size); f_mid_point = (float)i_mid_point / (float)i_model_unit; f_step_size = (float)i_step_size/(float)i_model_unit; f_spec_invq[bin] = MmseLaplace(f_mid_point, f_step_size); } } } }
NOTE: POST_GAIN_LUT[ ] is defined in table C.2 and STEP_SIZES_Q4_15[ ] in table C.11. GetRandomNoiseValue() is defined in clause 5.2.8.3. nRndState is the state of the random noise generator.
The used function MmseLaplace() is defined below.
Pseudocode 33
float MmseLaplace(float f_mid_point, float f_step_size) { f_upper = f_mid_point + f_step_size / 2.0f; f_lower = f_mid_point - f_step_size / 2.0f; f_pdf_lower = sqrt(2) / 2 * exp(-fabs(f_lower) * sqrt(2)); f_pdf_upper = sqrt(2) / 2 * exp(-fabs(f_upper) * sqrt(2)); // Prevent numerical problems if (f_pdf_lower < 0.0f) f_pdf_lower = 0.0f; if (f_pdf_upper < 0.0f) f_pdf_upper = 0.0f; f_mmse_n = f_pdf_lower*(sqrt(2)*f_lower-1)+f_pdf_upper*(sqrt(2)*f_upper+1); f_mmse_n /= sqrt(2)*(f_pdf_lower+f_pdf_upper)-2; if (f_lower > 0) { f_mmse_n = f_pdf_upper*(sqrt(2)*f_upper+1.0f); f_mmse_n -= f_pdf_lower*(sqrt(2)*f_lower+1.0f); f_mmse_n /= sqrt(2)*(f_pdf_upper-f_pdf_lower); } if (f_upper < 0) { f_mmse_n = f_pdf_upper*(sqrt(2)*f_upper-1.0f);
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)145
f_mmse_n -= f_pdf_lower*(sqrt(2)*f_lower-1.0f); f_mmse_n /= sqrt(2)*(f_pdf_upper-f_pdf_lower); } return f_mmse_n; }
5.2.5.2.5 Application of heuristic inverse scaling
Heuristic inverse scaling, as shown the following pseudocode, shall be applied if variance_preserving_flag is 0.
Pseudocode 34
// Heuristic inverse scaling for (band = 0; band < num_bands; band++) { f_gain_value = 1.0f / f_gain_q[band]; for (bin = start_bin[band] ; bin <= end_bin[band]; bin++) { f_spec_res[bin] = f_spec_invq[bin] * f_gain_value; } }
5.2.6 Subband predictor
Data and Control Interfaces
The input to the subband predictor is a vector of SSF output spectral lines from the previous block, a signal envelope vector for the current block, and the predictor parameters from the predictor decoder for the current block:
f_spec[bin] Vector of num_bins spectral lines from the previous block.
f_env_signal[band]
Vector of num_bands signal envelope values (float) for the current block. f_pred_gain
Predictor gain value for the current block. f_pred_lag
Predictor lag value for the current block.
The output of the subband predictor is a vector of predicted spectral lines:
f_spec_pred[bin] Vector of num_bins predicted spectral lines for the current block.
The bitstream and helper parameters used by the subband predictor are:
num_bands Number of bands.
n_mdct SSF block length.
num_bins Number of coded spectral lines.
The constants used by the subband predictor are:
NUM_SPEC_BUF = 5 Number of buffered spectra.
NUM_ENV_BUF = 4
Number of buffered envelopes.
The state internal to the subband predictor is the subband buffer and the envelope buffer:
f_spec_buffer[buf][bin] NUM_SPEC_BUF × num_bins dimensional array of spectral lines (float).
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)146
f_env_buffer[buf][band] NUM_ENV_BUF × num_bands dimensional array of signal envelope values (float).
The following intermediate signal is used by the subband predictor:
f_spec_extract[bin] Vector of num_bins spectral lines.
The input signals are stored in the internal subband and envelope buffer.
Pseudocode 35
// update of subband and envelope buffer for (spec = NUM_SPEC_BUF-1; spec > 0; spec--) { f_spec_buffer[spec] = f_spec_buffer[spec-1]; } f_spec_buffer[0] = f_spec[]; for (env = NUM_ENV_BUF-1; env > 0; env--) { f_env_buffer[env] = f_env_buffer[env-1]; } f_env_buffer[0] = f_env_signal[];
The extractor used by the subband predictor is a model based extractor. The period T0 in units of the MDCT block length is defined by:
T� =f_pred_lag
n_mdct.
An integer valued shift ks is set to:
ks = !0, T� ≤ 81/32;
1, T� > 81/32.
The reduced period is then given by T = T0 - ks and a table index nT is derived as follows:
nT = ! 0, T≤9/32;�16" + 0,5� − 4, T>9/32.
This index is used to select radii in time and frequency from table C.4 and table C.5:
Rt = TAB_PRED_RTS_TABLE�nT�, Rf=TAB_PRED_RFS_TABLE[nT].
A three-dimensional array of prediction coefficients of size (2Rf + 1)×65×Rt is selected using nT:
C�ν,η,k� = Call[nT]
NOTE 1: The setup of Call[ ] is described in clause 5.2.8.1.
From the subband buffer, an input array is created as follows:
Z�n,k� = f_spec_buffer[� + ks][�], � = 0,… ,# − 1,� = 0,… ,���_��� − 1
This input array is extended by zeros for n = num_bins,…,num_bins - 1 + Rf and by using even reflection Z�n,k� =
Z(−1 − �, �) for the negative values n = Rf,…,1.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)147
With these extensions, the extracted output is defined by the summation:
f_spec_extract[p]= $ % �−1��k+1�pC�ν,f�2� + & + 1�, ��Z(� + &, �)Rf
ν=-Rf
, � = 0,… ,���_��� − 1
Rt-1
k=0
Here, the function f is a composition of two functions f(µ)=g(φ(µ)), where the inner function is given by:
φ�µ� = 'T4
µ+0,5( − T
4µ,
and the outer function is given by:
g�φ� = ) 32, φ>T−32, φ<-T' 64*����2", 1� + 0,5( , otherwise.
The operation of the extractor is given in the following pseudocode.
Pseudocode 36
// model based prediction extraction f_period = f_pred_lag / n_mdct; // T_0 k_s = 0; if (f_period > 81.0/32.0) { k_s = 1; f_period -= 1.0f; // T (reduced period) } tab_idx = (f_period <= 9.0/32.0) ? 0 : floor(16*f_period + 0.5) - 4; // n_T Rt = PRED_RTS_TABLE[tab_idx]; Rf = PRED_RFS_TABLE[tab_idx]; C = C_all[tab_idx]; // create input array Z from subband buffer for (k = 0; k < Rt; k++) { for (n = 0; n < num_bins; n++) { Z[n][k] = f_spec_buffer[k+k_s][n]; } // extend Z for (n = num_bins; n < num_bins + Rf; n++) { Z[n][k] = 0.0f; } for (n = -Rf; n < 0; n++) { Z[n][k] = Z[-n-1][k]; } } // calculate extracted output for (bin = 0; bin < num_bins; bin++) { tmp = 0; for (nu = -Rf; nu <= Rf; nu++) { // calc f=f(2*bin+nu+1) mu = 2*bin+nu+1; phi = (f_period/4)*mu; phi = round(phi)-phi; if (phi > f_period) { f = 32; } else if (phi < -f_period) { f = -32; } else { min_2T = 2*f_period < 1 ? 2*f_period : 1; f = round(64*phi/min_2T); } for (k = 0; k < Rt; k++) { s = (bin % 2) ? s*(-1) : 1;
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)148
tmp += s*C[nu][f][k]*Z[bin+nu][k]; } } f_spec_extract[bin] = tmp; }
NOTE 2: PRED_RTS_TABLE[ ] and PRED_RFS_TABLE[ ] are defined in annex C, table C.5, and table C.4. The initialization of C_all[tab_idx] is defined in clause 5.2.8.1.
The operation of the shaper and the application of the prediction gain is specified in the following pseudocode.
Pseudocode 37
// shaper and prediction gain (for block index block) integer_lag = round(f_pred_lag/n_mdct); if ((b_iframe == 1) && (integer_lag > 0)) { // limit integer_lag to only use available envelop buffer entries integer_lag = 0; } for (band = 0; band < num_bands; band++) { f_envelope = 1.0f / f_env_buffer[integer_lag][band]; for (bin = start_bin[band] ; bin <= end_bin[band]; bin++) { f_spec_pred[bin] = f_spec_extract[bin] * f_envelope * f_pred_gain; } }
5.2.7 Inverse flattening
Data and control interfaces
The input to the inverse flattening is the spectrum from the spectrum decoder, the spectrum from the subband predictor, and the signal envelope values for the current block:
f_spec_res[bin] Vector of num_bins residual spectral lines from the spectrum decoder.
f_spec_pred[bin]
Vector of num_bins predicted spectral lines from the subband predictor. f_env_signal[band]
Vector of num_bands signal envelope values (float) for the current block.
The output from the inverse flattening is a vector of spectral lines:
sSSF,ch Vector of num_bins spectral lines for the current block of track ch.
The bitstream and helper parameters used by the inverse flattening are:
num_bands Number of bands.
num_bins
Number of coded spectral lines.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)149
Pseudocode 38
// Inverse flattening // compose f_spec_flat[bin] for (bin = 0; bin < num_bins; bin++) { f_spec_flat[bin] = f_spec_res[bin] + f_spec_pred[bin]; } // apply signal envelope for (band = 0; band < num_bands; band++) { for (bin = start_bin[band] ; bin <= end_bin[band]; bin++) { f_spec[bin] = f_spec_flat[bin] * f_env_signal[band]; } }
5.2.8 Parameterization
5.2.8.1 C matrix
The extractor of the subband predictor uses a matrix C, selected by tab_idx. The setup of all 37 possible matrices, stored in a big table C_ALL[tab_idx], is described in the following pseudocode. Each matrix C is a three-dimensional array of prediction coefficients of the size (2Rf + 1)×65×Rt, where Rt and Rf are dependent on tab_idx:
Rt=PRED_RTS_TABLE[tab_idx],
Rf=PRED_RFS_TABLE[tab_idx].
For the clarity of the description, C is addressed here with both positive and negative indices:
C�ν,η,k�, v=-Rf,…,Rf, η=-32,…,32, k=0,…,Rt-1
First, the sub-array for positive η indices is initialized with the array of quantized predictor coefficients given by table PRED_COEFF_QUANT_MAT[tab_idx], which is of size (2Rf + 1)×33×Rt. Next, the quantized prediction coefficients are replaced by the reconstructed values using the following formula for the reconstruction:
coeffreconstruct = 1,1787855 ×coeffquantized − 146
128
The remaining coefficients of the prediction coefficients array are initialized using the rule:
C�ν,η,k� = �−1�k+1C(−&, −,, �) For negative values, η = 32,…,1. (ν = Rf,…,Rf, k = 0,…,Rt - 1)
The setup is written in the following pseudocode.
Pseudocode 39
// setup of prediction coefficients arrays for (tab_idx = 0; tab_idx <= 36; tab_idx++) { Rf = PRED_RFS_TABLE[tab_idx]; Rt = PRED_RTS_TABLE[tab_idx]; // extract and reconstruct prediction coefficients for positive eta indices for (k = 0; k < Rt; k++) { for (eta = 0; eta < 33; eta++) { for (nu = -Rf; nu <= Rf; nu++) { C_ALL[tab_idx][nu][eta][k] = 1.1787855*(PRED_COEFF_QUANT_MAT[tab_idx][nu][eta][k] - 146)/128; } } } // initialize prediction coefficients for negative eta indices s = 1; for (k = 0; k < Rt; k++) { s *= -1; for (eta = -32; eta < 0; eta++) { for (nu = -Rf; nu <= Rf; nu++) { C_ALL[tab_idx][nu][eta][k] = s * C_ALL[tab_idx][-nu][-eta][k];
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)150
} } } }
5.2.8.2 Arithmetic coding
A single instance of an arithmetic coder is used during the arithmetic decoding of the following three parameter types within an SSF granule, each using a different statistical model:
• envelope indices using a static pre-trained CDF table;
• predictor gain indices using a static pre-trained CDF table;
• transform coefficients using CDF computation based on pre-trained CDF prototype function.
The arithmetic coder is initialized once per SSF granule, when the parsing has reached the ssf_ac_data element. The envelope quantization indices are coded first and a predefined look-up table (LUT) is used to obtain values of the cumulative distribution function (CDF) for respective indices. Next, the arithmetic coder operates in a loop of num_blocks blocks, where each block comprises coding of a single predictor gain index and a single set of quantization indices representing quantized MDCT coefficients. Depending on the parameter that is coded at a given time instance, the arithmetic coder uses either a LUT for the predictor gains, or computes the CDF values given dither realization and quantization step-sizes. The arithmetic decoding process is terminated only at the end of the SSF granule.
Termination bits are computed at the end of the decoding process so that an arithmetic decoder is able to provide the exact number of arithmetic decoded bits.
In the case of an I-frame, two envelopes are present instead of a single envelope.
Arithmetic coding implementation
The arithmetic decoding used within the SSF is specified using fixed point arithmetic and the two integer data types uint32 and int32. The unsigned type is used to represent the state of the arithmetic decoder and the signed type is used for counters in programme loops. The fixed point operators are defined below.
Pseudocode 40
// data types used by the arithmetic decoder typedef unsigned int uint32; /* ui */ typedef int int32; /* i */ // fixed point operators used by the arithmetic decoder // subtraction with conversion to signed variables #define SSF_I32_SUB_I32(a, b) ((int32)(a) - (int32)(b)) // addition with conversion to signed variables #define SSF_I32_ADD_I32(a, b) ((int32)(a) + (int32)(b)) // subtraction with a forced conversion to unsigned variables #define SSF_UI32_SUB_UI32(a, b) ((uint32)(a) - (uint32)(b)) // addition with a forced conversion to unsigned variables #define SSF_UI32_ADD_UI32(a, b) ((uint32)(a) + (uint32)(b)) // absolute value #define SSF_I32_ABS(a) ((a)>=0 ? a : -a) // multiplication in a 32 bit signed register #define SSF_I16_MUL_I16(a, b) ((int32)(a) * (int32)(b)) // multiplication in a 32 bit unsigned register #define SSF_UI16_MUL_UI16(a, b) ((uint32)(a) * (uint32)(b)) // left shift with a conversion to an unsigned integer #define SSF_UI32_SHIFT_LEFT(a, b) (((uint32)(a)) << (b)) // right shift with a conversion to an unsigned integer #define SSF_UI32_SHIFT_RIGHT(a, b) ((uint32)(a) >> (b))
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)151
// left shift of a signed integer (sign preserving) #define SSF_I32_SHIFT_LEFT(a, b) ((a) >= 0 ? (int32)SSF_UI32_SHIFT_LEFT(a, b) : -(int32)SSF_UI32_SHIFT_LEFT(-(a), b)) // right shift of a signed integer (sign preserving) #define SSF_I32_SHIFT_RIGHT(a, b) ((a) >= 0 ? (int32)SSF_UI32_SHIFT_RIGHT(a, b) : -(int32)SSF_UI32_SHIFT_RIGHT(-(a), b))
NOTE 1: Not all operators are actually used in the implementation.
NOTE 2: All variables involved in arithmetic coding are actually unsigned integers.
The implementation of the arithmetic decoder uses the following constant values.
Pseudocode 41
// constants used by the arithmetic decoder /* Number of model bits */ #define SSF_MODEL_BITS 15 /* All CDFs come in Q0.15 */ /* Model unit for the CDF specification */ #define SSF_MODEL_UNIT (1U<<(SSF_MODEL_BITS)) /* Number of range bits */ #define SSF_RANGE_BITS 30 /* Half of the range unit */ #define SSF_SSF_THRESHOLD_LARGE (1U<<((SSF_RANGE_BITS)-1)) /* Quarter of the range unit */ #define SSF_THRESHOLD_SMALL (1U<<((SSF_RANGE_BITS)-2)) /* Offset bits */ #define SSF_OFFSET_BITS 14
The following state structure represents the state of the arithmetic decoder.
Pseudocode 42
// state of the arithmetic decoder typedef struct _ac_state { // state uint32 uiLow; uint32 uiRange; uint32 uiOffset; uint32 uiOffset2; // fixed scales (updated during initialization - remain constant during operation) uint32 uiThresholdSmall; uint32 uiThresholdLarge; uint32 uiModelUnit; // specification - these values are also constant uint32 uiRangeBits; uint32 uiModelBits; // bitstream bitstream *psBs; // current position in the bitstream is stored internally } ac_state;
The initialization of the arithmetic decoder sets up the variables representing the state of the arithmetic decoder and reads the beginning of the arithmetic coded data, i.e. ssf_ac_data.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)152
Pseudocode 43
// initialize the arithmetic decoder int32 AcDecoderInit(ac_state *psS, // input & output bitstream *psBs) // input (pointer to the bitstream structure) { uint32 uiTmp; uint32 iBitIdx; /* Constant values */ psS->uiModelBits = SSF_MODEL_BITS; psS->uiModelUnit = SSF_MODEL_UNIT; psS->uiRangeBits = SSF_RANGE_BITS; psS->uiThresholdLarge = SSF_THRESHOLD_LARGE; psS->uiThresholdSmall = SSF_THRESHOLD_SMALL; /* Initialization */ psS->uiLow = 0; psS->uiRange = SSF_THRESHOLD_LARGE; /* Bitstream initialization */ psS->psBs = psBs; // just a pointer to some bitstream structure /* Read bits (Here we read SSF_RANGE_BITS from the bitstream.)*/ psS->uiOffset = (uint32)BsReadBit(psS->psBs); for (iBitIdx = 1; iBitIdx < psS->uiRangeBits; iBitIdx++) { uiTmp = (uint32)BsReadBit(psS->psBs); psS->uiOffset = SSF_UI32_SHIFT_LEFT(psS->uiOffset, 1); psS->uiOffset = SSF_UI32_ADD_UI32(psS->uiOffset, uiTmp); } psS->uiOffset2 = psS->uiOffset; return 0; }
Decoding a single symbol involves the usage of the following three functions. A function that decodes a single symbol given two CDF values is called AcDecodeSymbolExtCdf(). The function returns a signed quantization index and updates the state of the arithmetic decoder.
Pseudocode 44
int32 AcDecodeSymbolExtCdf(ac_state *psS, // input & output uint32 *puiCdfTable, // input const int32 iMinSymbol, // input const int32 iMaxSymbol) // input { // returns a decoded index uint32 uiTarget, uiCdfLow, uiCdfHigh, uiVal; int32 iFinalSymbol, iSymbolIdx; iFinalSymbol = (1U<<15); /* First we call AcDecodeTarget(). This is done once per symbol. */ uiTarget = AcDecodeTarget(psS); /* Q0.15 */ /* ===loop over the whole codebook=== */ /* Now we search over the entire codebook by computing the CDF values */ /* corresponding to the possible quantization indices. */ /* Assume that we have a codebook where the possible symbols are indexed */ /* starting form iMinSymbol and ending at iMaxSymbol. */ for (iSymbolIdx = iMinSymbol; iSymbolIdx <= iMaxSymbol; iSymbolIdx++) { if (puiCdfTable != NULL) { /* This is done when decoding the envelope indices and the predictor */ /* gain indices */ uiCdfLow = puiCdfTable[iSymbolIdx-iMinSymbol]; uiCdfHigh = puiCdfTable[iSymbolIdx-iMinSymbol+1]; } else { /* For the transform coefficients we use the actual CDF computation. */ /* See code below, how to derive uiCdfLow and uiCdfHigh from */ /* iSymbolIdx, i_step_size and i_dither_val */ } /* Now, if uiTarget is between uiCdfHigh and uiCdfLow we found the symbol. */ if ((uiTarget < uiCdfHigh) && (uiTarget >= uiCdfLow)) { iFinalSymbol = iSymbolIdx; /* we have just found a symbol index */ /* Let the arithmetic decoder advance in the bitstream and */
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)153
/* update its state */ AcDecode(uiCdfLow, uiCdfHigh, psS); break; /* we are done - quit the codebook search loop */ } } /* end of the loop over the whole codebook */ return iFinalSymbol; }
Pseudocode 45
int32 AcDecodeTarget(ac_state *psS) // input & output { uint32 uiTarget, uiRange, uiNum, uiDen, uiTmp, uiNumShifts; int32 iIdx; uiRange = SSF_UI32_SHIFT_RIGHT(psS->uiRange, psS->uiModelBits); uiTmp = SSF_UI32_SHIFT_LEFT(1, SSF_OFFSET_BITS); if (uiRange < uiTmp) { uiNumShifts = psS->uiModelBits; } else { uiNumShifts = psS->uiModelBits - 1; } uiNum = psS->uiOffset; uiDen = SSF_UI32_SHIFT_LEFT(uiRange, uiNumShifts); uiTarget = 0; // initialize for (iIdx = uiNumShifts; iIdx > 0; iIdx--) { if (uiNum >= uiDen) { uiNum = SSF_UI32_SUB_UI32(uiNum, uiDen); uiTarget = SSF_UI32_ADD_UI32(uiTarget, 1); } uiNum = SSF_UI32_SHIFT_LEFT(uiNum, 1); uiTarget = SSF_UI32_SHIFT_LEFT(uiTarget, 1); } if (uiNum >= uiDen) { uiNum = SSF_UI32_SUB_UI32(uiNum, uiDen); uiTarget = SSF_UI32_ADD_UI32(uiTarget, 1); } if (uiTarget >= psS->uiModelUnit) { uiTarget = SSF_UI32_SUB_UI32(psS->uiModelUnit, 1); } return uiTarget; /* Q0.15 by design */ }
Pseudocode 46
int32 AcDecode(uint32 uiCdfLow, // input uint32 uiCdfHigh, // input ac_state *psS) // input & output { uint32 uiTmp1, uiTmp2; uint32 uiRange; uiRange = SSF_UI32_SHIFT_RIGHT(psS->uiRange, psS->uiModelBits); uiTmp1 = SSF_UI16_MUL_UI16(uiRange, uiCdfLow); psS->uiOffset = SSF_UI32_SUB_UI32(psS->uiOffset, uiTmp1); if (uiCdfHigh < psS->uiModelUnit) { uiTmp2 = SSF_UI32_SUB_UI32(uiCdfHigh, uiCdfLow); psS->uiRange = SSF_UI16_MUL_UI16(uiRange, uiTmp2); } else { psS->uiRange = SSF_UI32_SUB_UI32(psS->uiRange, uiTmp1); } // denormalize while (psS->uiRange <= psS->uiThresholdSmall)
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)154
{ /* Read a single bit from the bitstream */ uiTmp1 = (uint32)BsReadBit(psS->psBs); psS->uiRange = SSF_UI32_SHIFT_LEFT(psS->uiRange, 1); psS->uiOffset = SSF_UI32_SHIFT_LEFT(psS->uiOffset, 1); psS->uiOffset = SSF_UI32_ADD_UI32(psS->uiOffset, uiTmp1); psS->uiOffset2 = SSF_UI32_SHIFT_LEFT(psS->uiOffset2, 1); if (psS->uiOffset & 1) { psS->uiOffset2++; } } return 0; }
Computing the number of termination bits happens once per granule. The AcDecodeFinish() function is needed so the decoder can tell how many bits have been successfully decoded from the bitstream.
Pseudocode 47
int32 AcDecodeFinish(ac_state *psS) // input & output { int32 iRes, iBitIdx; uint32 uiConstUpFact, uiBits, uiVal; uint32 uiTmp1, uiTmp2, uiRevIdx; iRes = psS->psBs->iPos; // iPos: bits read so far by the bitstream reader iRes = iRes - psS->uiRangeBits; /* Determine the finish bits */ psS->uiLow = (psS->uiOffset2 & (psS->uiThresholdLarge-1)); uiTmp1 = SSF_UI32_SUB_UI32(psS->uiThresholdLarge, psS->uiOffset); psS->uiLow = SSF_UI32_ADD_UI32(psS->uiLow, uiTmp1); for (iBitIdx = 1; iBitIdx <= psS->uiRangeBits; iBitIdx++) { uiRevIdx = psS->uiRangeBits - iBitIdx; uiConstUpFact = SSF_UI32_SHIFT_LEFT(1U, uiRevIdx); uiConstUpFact = SSF_UI32_SUB_UI32(uiConstUpFact, 1U); uiTmp1 = SSF_UI32_ADD_UI32(psS->uiLow, uiConstUpFact); uiBits = SSF_UI32_SHIFT_RIGHT(uiTmp1, uiRevIdx); uiVal = SSF_UI32_SHIFT_LEFT(uiBits, uiRevIdx); uiTmp1 = SSF_UI32_ADD_UI32(uiVal, uiConstUpFact); uiTmp2 = SSF_UI32_SUB_UI32(psS->uiRange, 1U); uiTmp2 = SSF_UI32_ADD_UI32(uiTmp2, psS->uiLow); if ((psS->uiLow <= uiVal) && (uiTmp1 <= uiTmp2)) { break; } } iRes = iRes + iBitIdx; return iRes; // returns number of arithmetically decoded bits }
The arithmetic decoding of the envelope indices using the above mentioned function AcDecodeSymbolExtCdf() is given by the following pseudocode
Pseudocode 48
// envelope indices AC decoding for (idx = 1; idx < num_bands; idx++) { env_idx[idx] = AcDecodeSymbolExtCdf(ac_state, ENVELOPE_CDF_LUT, 0, 32); }
NOTE 3: ENVELOPE_CDF_LUT is defined in table C.8.
Similarly, the arithmetic decoding of the predictor gain indices can be described as follows.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)155
Pseudocode 49
// predictor gain indices AC decoding arithmetic_decode_pred() { i_pred_gain_idx = AcDecodeSymbolExtCdf(ac_state, PREDICTOR_GAIN_CDF_LUT, 0, 32); return i_pred_gain_idx; }
NOTE 4: PREDICTOR_GAIN_CDF_LUT is defined in table C.7.
The arithmetic decoding of the transform coefficients can be done according to the following pseudocode.
Pseudocode 50
// transform coefficients AC decoding // input: i_alloc_table[band], i_dither[bin] // output: i_quant_idx[bin] arithmetic_decode_coeffs() { for(band = 0; band < num_bands; band++) { i_alloc = i_alloc_table[band]; if (i_alloc == 0) { for (bin = start_bin[band]; bin <= end_bin[band]; bin++) { i_quant_idx[bin] = 0; } } else // (i_alloc > 0) { i_step_size = STEP_SIZES_Q4_15[i_alloc]; i_max_idx = AC_COEFF_MAX_INDEX[i_alloc] + 1; for (bin = start_bin[band]; bin <= end_bin[band]; bin++) { i_dither_val = i_dither[bin]; i_quant_idx[bin] = AcDecodeSymbolExtCdf(ac_state, NULL, 0, i_max_idx); } } } }
NOTE 5: AC_COEFF_MAX_INDEX[ ] is defined in table C.12.
The function AcDecodeSymbolExtCdf() is called without a CDF table to signal that the CDF needs to be calculated.
NOTE 6: For the CDF calculation, the values i_step_size and i_dither_val are needed by AcDecodeSymbolExtCdf(). These values are also passed on to this function although they are not shown as input parameters.
The calculation of the CDF is done according to the following pseudocode.
Pseudocode 51
// CDF calculation to be used during the AC decoding of the transform coefficients // uiCdfLow and uiCdfHigh are derived from iSymbolIdx, i_step_size and i_dither_val const int32 i_max_value = 327680; int32 iLeft; int32 iRight; int32 iMidpoint; int32 i_half_step_size; iMidpoint = Idx2Reconstruction(iSymbolIdx, i_dither_val, i_step_size); i_half_step_size = SSF_I32_SHIFT_RIGHT(i_step_size, 1); iLeft = SSF_I32_SUB_I32(iMidpoint, i_half_step_size);
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)156
iRight = SSF_I32_ADD_I32(iLeft, i_step_size); iLeft = iLeft < -i_max_value ? -i_max_value : iLeft; iRight = iRight > i_max_value ? i_max_value : iRight; uiCdfLow = CdfEst(iLeft); uiCdfHigh = CdfEst(iRight);
The used functions Idx2Reconstruction() and CdfEst() are given in the following pseudocode examples.
Pseudocode 52
int32 Idx2Reconstruction(int32 iIndex, int32 iDitherValue, int32 iStepSize) { int32 iReconstruction, iTmp1, iTmp2; // subtract the dither iTmp1 = SSF_I32_SHIFT_LEFT(iIndex, 15); iReconstruction = SSF_I32_SUB_I32(iTmp1, iDitherValue); // multiply times the step size iTmp2 = SSF_I32_SHIFT_RIGHT(iReconstruction, 15); iTmp1 = SSF_I32_SHIFT_LEFT(iTmp2, 15); iTmp1 = SSF_I32_SUB_I32(iReconstruction, iTmp1); iTmp1 = SSF_I32_SHIFT_RIGHT(iTmp1, 3); iReconstruction = SSF_I16_MUL_I16(iTmp1, iStepSize); iReconstruction = SSF_I32_SHIFT_RIGHT(iReconstruction, 12); iTmp1 = SSF_I16_MUL_I16(iTmp2,iStepSize); iReconstruction = SSF_I32_ADD_I32(iReconstruction, iTmp1); return iReconstruction; }
Pseudocode 53
uint32 CdfEst(int32 iInVal) // input value Qx.15 { int32 iIdx; // Q4.15->Q.0 (we keep 9 MSB, result in [-352, 352] for a correctly bounded input) iIdx = SSF_I32_SHIFT_RIGHT(iInVal, 10); // note that it is a shift with a sign // iIdx is an offset from the beginning of the table [-352, 352] iIdx = iIdx + 352; return CDF_TABLE[iIdx]; // read from the CDF table }
NOTE 7: CDF_TABLE[ ] is defined in table C.6.
5.2.8.3 Dither and random noise
Two instances of a random number generator are used during the SSF decoding: one for the dither signal setup and the other one for noise signal values. The implementation of the random number generator shall conform to the code in the following pseudocode since it is important that the SSF encoder and the SSF decoder use the same pseudo-random values. The random number generators are initialized at the beginning of the decoding of an SSF-I-frame.
The following state structure represents the state of the random number generator.
Pseudocode 54
// This structure is used to store the state of the random number generator. // All variables in the state structure are subject to implicit modulo 256. typedef struct _ssf_rndgen_state { uint8 uiOffsetA; // scan step size uint8 uiOffsetB; // scan offset uint8 uiStateIdx; // state counter 1 uint8 uiCurrentIdx; // state counter 2 (main index)
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)157
} ssf_rndgen_state;
The initialization of the random number generator sets up the variables representing the state of the random number generator.
Pseudocode 55
// initialize / reset the random number generator void ResetRandGenState(ssf_rndgen_state *psS) { psS->uiOffsetA = 0; psS->uiOffsetB = 0; psS->uiStateIdx = 1; psS->uiCurrentIdx = 0; }
The following function, which returns a random value, is used during the setup of the dither signal.
Pseudocode 56
int32 GetDitherValue(ssf_rndgen_state *psS) { int32 iRes; // main look-up iRes = DITHER_TABLE[psS->uiCurrentIdx]; // state update (using implicit modulo 256) psS->uiOffsetA++; psS->uiStateIdx = psS->uiStateIdx++; if (psS->uiOffsetA == 255) { psS->uiCurrentIdx = psS->uiCurrentIdx++; psS->uiOffsetB++; psS->uiOffsetA = 0; } psS->uiCurrentIdx = psS->uiCurrentIdx + psS->uiOffsetA; psS->uiStateIdx = psS->uiStateIdx + psS->uiOffsetB + psS->uiOffsetA; return iRes; }
The following function, which returns a random value, is used by the spectrum decoder to insert noise values.
Pseudocode 57
float32 GetRandomNoiseValue(ssf_rndgen_state *psS) { float32 fRes, fRes1, fRes2; // main look-up fRes1 = RANDOM_NOISE_TABLE[psS->uiCurrentIdx]; fRes2 = RANDOM_NOISE_TABLE[psS->uiStateIdx]; // compute result fRes = fRes1 + fRes2; // state update (using implicit modulo 256) psS->uiOffsetA++; psS->uiStateIdx=psS->uiStateIdx++; if (psS->uiOffsetA == 255) { psS->uiCurrentIdx = psS->uiCurrentIdx++; psS->uiOffsetB++; psS->uiOffsetA = 0; } psS->uiCurrentIdx = psS->uiCurrentIdx + psS->uiOffsetA; psS->uiStateIdx = psS->uiStateIdx + psS->uiOffsetB + psS->uiOffsetA;
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)158
return fRes; }
To generate the dither signal to be used by the decoding of the current SSF granule, the above introduced function GetDitherValue() is employed.
Pseudocode 58
// setup of i_dither_cur[block][bin] // ditherGenState holds the state of the random number generator used for the dither // signal generation for (block = 0; block < num_blocks; block++) { for (bin = 0; bin < num_bins; bin++) { i_dither_cur[block][bin] = GetDitherValue(&ditherGenState); } }
5.3 Stereo and multichannel processing
5.3.1 Introduction
The multichannel tools take their input from one or more spectral frontends, receiving n-tuples of spectra with matching time/frequency layout. The multichannel processing tool applies a number time-varying and frequency-varying matrix operation on these tuples. Between two and seven spectra are transformed at a time.
Parameters for the transformations are transmitted by chparam_info() elements; the various transform matrices M (up to 7 x 7) are built up from these parameters as described in the following paragraphs.
When the tuples have been transformed, they are passed on to the IMDCT, defined in clause 5.5. While the input to the multichannel processing tool does not have a "channel" meaning, the output from the tool carries channels, ordered as L, R, C, Ls, Rs, etc.
Data and Control Interfaces
The input to the stereo and multichannel processing tool are scaled spectral lines of tracks derived from decoding the sf_data element stored in one of the following elements using the ASF or the SSF tool:
• channel_pair_element (MSAP = 2),
• 3_0_channel_element (MSAP = 3),
• 5_X_channel_element (MSAP = 5), or
• 7_X_channel_element (MSAP = 7).
s SMP,[0|1|…] Up to MSAP vectors of spectral lines, each vector representing a track decoded from an sf_data element.
NOTE 1: The tracks are numbered according to their occurrence in the bitstream, starting from track sSMP,0.
The output from the stereo and multichannel processing tool are M SAP blocks of scaled spectral lines:
s SMP,[L|R|C|…] MSAP vectors of blk_len spectral lines assigned to channels with discrete speaker locations.
NOTE 2: See clause D.1 for a listing of channel abbreviations.
The bitstream and additional information used by the stereo audio processing tool is:
blk_len Block length. Equals to the number of input and output spectral lines in one channel.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)159
N Block length. Equals to the number of input and output spectral lines in one channel.
sap_used[g][b] Array indicating the operating mode of the stereo audio processing tool for group g and scale factor band sfb.
sap_gain[g][b] Array of real-valued gains for group g and scale factor band sfb.
5.3.2 Parameter extraction
All parameters for the multichannel tool are transmitted in one or more chparam_info() elements. The following pseudocode describes parsing the chparam_info() element and extracting parameters a, b, c, d.
Pseudocode 59
// initialize a[g][sfb], b[g][sfb], c[g][sfb] and d[g][sfb] after parsing of // chparam_info()
max_sfb_prev = get_max_sfb(0);
for (g = 0; g < num_window_groups; g++) { max_sfb_g = get_max_sfb(g); for (sfb = 0; sfb < max_sfb_g; sfb++) { if (sap_mode == 0 || (sap_mode == 1 && ms_used[g][sfb] == 0)) { a[g][sfb]=d[g][sfb]=1; b[g][sfb]=c[g][sfb]=0; } else if (sap_mode == 2 || ((sap_mode == 1 && ms_used[g][sfb] == 1)) { a[g][sfb]=b[g][sfb]=c[g][sfb]=1; d[g][sfb]=-1; }
if (sap_used[g][sfb]){
// inverse quantize alpha_q[g][sfb] sap_gain[g][sfb] = alpha_q[g][sfb] * 0.1f; a[g][sfb]=1+sap_gain[g][sfb]; b[g][sfb]=1; c[g][sfb]=1-sap_gain[g][sfb]; d[g][sfb]=-1;
} else { a[g][sfb]=1; b[g][sfb]=0; c[g][sfb]=0; d[g][sfb]=1; }
}
else { // sap_mode == 3 sap_used[g][sfb] = sap_coeff_used[g][sfb]; // setup alpha_q[g][sfb] if (sap_used[g][sfb]) { if (sfb % 2) { alpha_q[g][sfb] = alpha_q[g][sfb-1]; } else { delta = dpcm_alpha_q[g][sfb] - 60; if ((g == 0) || (max_sfb_g != max_sfb_prev)) { code_delta = 0; }
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)160
else { code_delta = delta_code_time; } if (code_delta) { alpha_q[g][sfb] = alpha_q[g-1][sfb] + delta; } else if (sfb == 0) { alpha_q[g][sfb] = delta; } else { alpha_q[g][sfb] = alpha_q[g][sfb-2] + delta; } } // inverse quantize alpha_q[g][sfb] sap_gain[g][sfb] = alpha_q[g][sfb] * 0.1f; a[g][sfb]=1+sap_gain[g][sfb]; b[g][sfb]=1; c[g][sfb]=1-sap_gain[g][sfb]; d[g][sfb]=-1; } else { a[g][sfb]=1; b[g][sfb]=0; c[g][sfb]=0; d[g][sfb]=1; } }
}
max_sfb_prev = max_sfb_g;
}
5.3.3 Processing the channel data elements
5.3.3.0 Introduction
In the various channel data elements, a number n > 1 of input tracks (I0, …, In) are matrix transformed into the n output tracks (O0, …, On). They share the same sf_info and so share the same time/frequency resolution. The matrix equations below are understood to apply to each time/frequency tile separately, with separate parameters a, b, c, d per such tile.
In clause 5.3.3.1 through clause 5.3.3.5, decoding an sf_data element is understood to mean decoding the spectral frontend data.
5.3.3.1 Processing tracks of the mono_data element
Let I0 be the track decoded from the sf_data element contained in the mono_data element.
The decoder shall map the input track directly to the output O0, i.e.: �O�� = �I�� 5.3.3.2 Processing tracks of the two_channel_data or stereo_data element
Let I0 and I1 be the tracks decoded from the two sf_data elements contained in the two_channel_data element, or of the stereo_data element.
If b_enable_mdct_stereo_proc is false, no further processing is necessary and O0=I0, O1=I1.
Otherwise, the decoder shall extract the parameters a, b, c, d from the contained chparam_info element as described in clause 5.3.2 and derive the outputs O0, O1 by:
-O�
O�. = /a b
c d0 × -I�I�.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)161
5.3.3.3 Processing tracks of the three_channel_data element
Let I0, I1, and I2 be the tracks decoded from the three sf_data elements contained in the three_channel_data element.
The decoder shall extract parameters a i, b i, c i, d i (i = 0…1) from the two contained chparam_info elements, and from the chel_matsel element determine the transform matrix M as described in table 177.
Finally, the output tracks O0, O1 and O2 shall be derived by:
1O�
O�
O�
2 = M × 1I�I�I�
2
Table 177: Determining the three_channel_data() element transform matrix
chel_matsel M 0 �a�a� b�a� b�
c� d� 0
a�c� b�c� d��
1 � d� c� 0
b�a� a�a� b�b�c� a�c� d�
� 2 �a�a� b� b�a�
a�c� d� b�c�c� 0 d�
� 3 �a� c�b� d�b�
0 a� b�c� c�d� d�d�
� 4 � a� 0 b�
c�b� a� d�b�c�d� c� d�d�
� 5 �a� d�b� c�b�
c� d�d� c�d�0 b� a�
� 6 �d�d� c�d� c�
b� a� 0
d�b� c�b� a��
7 � a� b� 0
c�d� d�d� c�c�b� d�b� a�
� 8 �d�d� c� c�d�
d�b� a� c�b�b� 0 a�
� 9 �d� b�c� a�c�
0 d� c�b� b�a� a�a�
� 10 � d� 0 c�
b�c� d� a�c�b�a� b� a�a�
� 11 �d� a�c� b�c�
b� a�a� b�a�0 c� d�
�
5.3.3.4 Processing tracks of the four_channel_data element
Let I0…I3 be the tracks decoded from the four sf_data elements contained in the four_channel_data element.
The decoder shall extract parameters ai, bi, ci, di (i = 0…3) from the four contained chparam_info elements.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)162
Finally, the output tracks O0… O3 shall be derived by:
3O�
O�
O�
O�
4 = 3a�a� b�a� a� b� b� b�c�a� d�a� c�b� d�b�a�c� b�c� a�d� b�d�c�c� d�c� c�d� d�d�
4 × 3I�I�I�I�
4
5.3.3.5 Processing tracks of the five_channel_data element
Let I0…I4 be the tracks decoded from the five sf_data elements contained in the five_channel_data element.
Next the decoder shall extract parameters ai, bi, ci, di (i = 0…4) from the five contained chparam_info elements, and from the chel_matsel element determine the transform matrix M as described in table 178.
Finally, the output tracks O0… O4 shall be derived by:
56667O�
O�
O�
O�
O�8999: = M ×
56667I�I�I�I�I�89
99:
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)163
Table 178: Determining the five_channel_data() element transform matrix
chel_matsel M 0
�����a�a�a� b�a�a� b�a� a�b� b�b�
c�a� d�a� 0 c�b� d�b�a�c� b�c� d� 0 0
a�a�c� b�a�c� b�c� a�d� b�d�c�c� d�c� 0 c�d� d�d���
���
1
����� d�a� c�a� 0 a�b� b�b�b�a�a� a�a�a� b�a� c�b� d�b�b�c� a�c� d� 0 0
d�c� c�c� 0 a�d� b�d�b�a�c� a�a�c� b�c� c�d� d�d���
���
2
�����a�a�a� b�a� b�a�a� a�b� b�b�a�c�a� d�a� b�c�a� c�b� d�b�
c� 0 d� 0 0
a�a�c� b�c� b�a�c� a�d� b�d�a�c�c� d�c� b�c�c� c�d� d�d���
���
3
�����a�a� c�b�a� d�b�a� a�b� b�b�0 a�a� b�a� c�b� d�b�c� c�d� d�d� 0 0
a�c� c�b�c� d�b�c� a�d� b�d�0 a�c� b�c� c�d� d�d���
���
4
����� a�a� 0 b�a� a�b� b�b�c�b�a� a�a� d�b�a� c�b� d�b�c�d� c� d�d� 0 0
a�c� 0 b�c� a�d� b�d�c�b�c� a�c� d�b�c� c�d� d�d���
���
5
�����a�a� d�b�a� c�b�a� a�b� b�b�c�a� d�d�a� c�d�a� c�b� d�b�0 b� a� 0 0
a�c� d�b�c� c�b�c� a�d� b�d�c�c� d�d�c� c�d�c� c�d� d�d���
���
6
�����d�d�a� c�d�a� c�a� a�b� b�b�
b�a� a�a� 0 c�b� d�b�d�b� c�b� a� 0 0
d�d�c� c�d�c� c�c� a�d� b�d�b�c� a�c� 0 c�d� d�d���
���
7
����� a�a� b�a� 0 a�b� b�b�c�d�a� d�d�a� c�a� c�b� d�b�c�b� d�b� a� 0 0
a�c� b�c� 0 a�d� b�d�c�d�c� d�d�c� c�c� c�d� d�d���
���
8
�����d�d�a� c�a� c�d�a� a�b� b�b�d�b�a� a�a� c�b�a� c�b� d�b�
b� 0 a� 0 0
d�d�c� c�c� c�d�c� a�d� b�d�d�b�c� a�c� c�b�c� c�d� d�d���
���
9
�����d�a� b�c�a� a�c�a� a�b� b�b�0 d�a� c�a� c�b� d�b�b� b�a� a�a� 0 0
d�c� b�c�c� a�c�c� a�d� b�d�0 d�c� c�c� c�d� d�d���
���
10
����� d�a� 0 c�a� a�b� b�b�b�c�a� d�a� a�c�a� c�b� d�b�b�a� b� a�a� 0 0
d�c� 0 c�c� a�d� b�d�b�c�c� d�c� a�c�c� c�d� d�d���
���
11
�����d�a� a�c�a� b�c�a� a�b� b�b�b�a� a�a�a� b�a�a� c�b� d�b�0 c� d� 0 0
d�c� a�c�c� b�c�c� a�d� b�d�b�c� a�a�c� b�a�c� c�d� d�d���
���
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)164
5.3.4 Processing the channel elements
5.3.4.0 Introduction
Tracks routed into the stereo and multiprocessing tool (sSMP,[�|�|�|… ]) may originate from four different channel elements: channel_pair_element, 3_0_channel_element, 5_X_channel_element or 7_X_channel_element.
Processing and mapping from input tracks to output channels (sSMP,[L|R|C|… ]) differs depending on the channel element, the codec mode and the coding config.
5.3.4.1 Processing tracks of a channel_pair_element
If stereo_codec_mode ∈ {SIMPLE,ASPX,ASPX_ACPL_1}, the 2 tracks of the stereo_data() element shall be processed according to clause 5.3.3.2. The mapping of input tracks and output channels shall be done as follows:
�I�I�� = �sSMP,�sSMP,�
� �sSMP,L
sSMP,R� = �O�
O��
If stereo_codec_mode = ASPX_ACPL_2, the single track of the sf_data() element shall be mapped to the output channels as follows:
�sSMP,L
sSMP,R� = �1 0��sSMP,��
5.3.4.2 Processing tracks of a 3_0_channel_element
If 3_0_coding_config = 0, the 2 tracks of the stereo_data() element (sSMP,0, sSMP,1) are processed according to clause 5.3.3.2. The mapping of input tracks and output channels shall be done as follows:
�I�I�� = �sSMP,�sSMP,�
� �sSMP,L
sSMP,R� = �O�
O��
The third track of the additional mono_data() element, sSMP,2, shall be processed according to clause 5.3.3.1. The mapping shall be done as follows: �I�� = �sSMP,�� �sSMP,C� = �O�� If 3_0_coding_config = 1, the three_channel_data() element is processed according to clause 5.3.3.3. The mapping of input and output tracks is done as follows:
I�I�I� = sSMP,�
sSMP,�sSMP,�
sSMP,L
sSMP,R
sSMP,C
= O�O�O�
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)165
5.3.4.3 Processing tracks of a 5_X_channel_element
5.3.4.3.0 Introduction
The up to 5 tracks acquired from a 5_X_channel_element shall be processed according to the channel data element they originate from. Tables in this clause describe which tracks shall be taken as input, and which channels are produced as output when processing the various channel data elements according to clause 5.3.3.0.
To improve readability, input and output vectors are denoted as row vectors. Furthermore, numbers in the input vector elements [0, 1, …] represent the input tracks [sSMP,0, sSMP,1, …] according to the bitstream order of the channel data elements. Letters in the output vector elements [L, R, …] represent output channels [sSMP,L, sSMP,R, …], respectively.
5.3.4.3.1 5_X_codec_mode ∈ {SIMPLE, ASPX}
If 5_X_codec_mode ∈ {SIMPLE, ASPX}, five audio tracks have been acquired from the bitstream. The mapping of input tracks and output channels is described in table 179. If coding_config = 0, the element 2ch_mode determines the final output channel mapping of the two two_channel_data() elements.
Table 179: Input and output mapping for 5_x_codec_mode ∈ {SIMPLE, ASPX}
coding_config 0 1 2 3 element Input Output Input Output Input Output Input Output
2ch_mode - 0 1 - - - - - - mono_data [4] [C] [C] - - [4] [C] - - 1st two_channel_data [0,1] [L,R] [L,Ls] [3,4] [Ls,Rs] - - - - 2nd two_channel_data [2,3] [Ls,Rs] [R,Rs] - - - - - - three_channel_data - - - [0,1,2] [L,R,C] - - - - four_channel_data - - - - - [0,1,2,3] [L,R,Ls,Rs] - - five_channel_data - - - - - - - [0,1,2,3,4] [L,R,C,Ls,Rs]
5.3.4.3.2 5_X_codec_mode ∈ {ASPX_ACPL_1,ASPX_ACPL_2}
If 5_X_codec_mode ∈ {ASPX_ACPL_1,ASPX_ACPL_2}, three audio tracks have been acquired from channel data elements. The mapping of input tracks and preliminary output channels [A, B, C] is described in table 180.
Table 180: Input and output mapping for 5_x_codec_mode ∈ {ASPX_ACPL_[1|2]}
coding_config 0 1 element Input Output Input Output
mono_data [2] [C] - - 1st two_channel_data [0,1] [A,B] - - three_channel_data - - [0,1,2] [A,B,C]
If 5_X_codec_mode = ASPX_ACPL_1, two additional audio tracks have been acquired from two single sf_data elements, denoted as [sSMP,3, sSMP,4]. Using the extracted parameters [a0, b0, c0, d0] and [a1, b1, c1, d1] from the two associated chparam_info elements as described in clause 5.3.2, the output channels can be generated according to:
���� sSMP,LsSMP,R
sSMP,C
sSMP,LssSMP,Rs���
��=
���� a� 0 0 b� 0
0 a� 0 0 b�0 0 1 0 0c� 0 0 d� 0
0 c� 0 0 d������ × ���
� sSMP,AsSMP,B
sSMP,C
sSMP,�sSMP,����
��
If 5_X_codec_mode = ASPX_ACPL_2, the output channel generation simplifies to:
���� sSMP,LsSMP,R
sSMP,C
sSMP,LssSMP,Rs���
��=
���� 1 0 00 1 0
0 0 1
0 0 00 0 0���
��× sSMP,A
sSMP,B
sSMP,C
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)166
5.3.4.3.3 5_X_codec_mode = ASPX_ACPL_3
If 5_X_codec_mode = ASPX_ACPL_3, the two audio tracks (sSMP,0, sSMP,1) are acquired from a stereo_data element and processed according to clause 5.3.3.2. The mapping of input tracks and output channels shall be done as follows:
�I�I�� = �sSMP,�sSMP,�
�
���� sSMP,LsSMP,R
sSMP,C
sSMP,LssSMP,Rs���
��=
���� 1 00 1
0 0
0 00 0���
��× �O�
O��
5.3.4.4 Processing tracks of the 7_X_channel_element
5.3.4.4.0 Introduction
The 7_X_channel_element enables the decoder for two more channels beyond L, R, C, Ls and Rs (refer to table 88 for details). Analogously to the 5_X_channel_element, up to 7 tracks of a 7_X_channel_element shall be processed according to the channel data element from which they originate. Table notations are borrowed from clause 5.3.4.3. The mapping is partially done to preliminary channel descriptors first, which are again mapped to the actual channels afterwards. The final channel mapping in table 182 should describe the following matrix multiplication process after the creation of the preliminary outputs described in table 181.
5.3.4.4.1 7_X_codec_mode ∈ {SIMPLE, ASPX}
If 7_X_codec_mode ∈ {SIMPLE,ASPX}, seven audio tracks are available. The mapping of input tracks and preliminary output channel descriptors (A,B,D,E,F,G) is described in table 181. The processing for each of the channel data elements shall be done according to clause 5.3.3.0.
NOTE: For coding_config = 0, the output channel mapping for the first two two_channel_data elements depends on the element 2ch_mode.
Table 181: Input and preliminary output mapping for 7_X_codec_mode ∈ {SIMPLE,ASPX}
coding_config 0 1 2 3 element Input Output Input Output Input Output Input Output
2ch_mode - 0 1 - - - - - - mono_data [6] [C] [C] - - [6] [C] - - 1st two_channel_data [0,1] [A,B] [A,D] [3,4] [D,E] [4,5] [F,G] [5,6] [F,G] 2nd two_channel_data [2,3] [D,E] [B,E] [5,6] [F,G] - - - - 3rd two_channel_data [4,5] [F,G] [F,G] - - - - - - three_channel_data - - - [0,1,2] [A,B,C] - - - - four_channel_data - - - - - [0,1,2,3] [A,B,D,E] - - five_channel_data - - - - - - - [0,1,2,3,4] [A,B,C,D,E]
If b_use_sap_add_ch is true, the decoder shall extract parameters ai, bi, ci, di (i = 0…1) from the two contained chparam_info elements. If b_use_sap_add_ch = false, the decoder shall assume the following parameters: ai = di = 1, bi = ci = 0. Table 182 outlines how the decoder output shall be constructed, by mapping the preliminary output channel descriptors to the actual channels.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)167
Table 182: Final channel mapping for 7_X_codec_mode ∈ {SIMPLE,ASPX}
channel_mode Out 3/4/0.x � Ls
LrsRsRrs
� = �a� b� 0 0
c� d� 0 0
0 0 a� b�0 0 c� d�
� × �DFEG
� �LR� = �A
B�
5/2/0.x � LLwR
Rw
� = �a� b� 0 0
c� d� 0 0
0 0 a� b�0 0 c� d�
� × �AFBG
� �LsRs
� = �DE�
3/2/2.x � LVhlR
Vhr
� = �a� b� 0 0
c� d� 0 0
0 0 a� b�0 0 c� d�
� × �AFBG
� �LsRs
� = �DE�
5.3.4.4.2 7_X_codec_mode = ASPX_ACPL_1
If 7_X_codec_mode = ASPX_ACPL_1, seven audio tracks are available. Processing is done analogously to the case when 7_X_codec_mode = SIMPLE and b_use_sap_add_ch is false, with the difference that the output channels F and G are derived from two sf_data elements and their associated chparam_info elements instead of the two_channel_data element.
5.3.4.4.3 7_X_codec_mode = ASPX_ACPL_2
If 7_X_codec_mode = ASPX_ACPL_2, five audio tracks are available. The mapping of input tracks and preliminary output channel descriptors (A,B,D,E,F,G) is described in table 183. The processing for each of the channel data elements shall be done according to clause 5.3.3.0.
Table 183: Input and preliminary output mapping for 7_X_codec_mode = ASPX_ACPL_2
coding_config 0 1 2 3 element Input Output Input Output Input Output Input Output
2ch_mode 0 1 mono_data [4] [C] [C] - - [4] [C] - - 1st two_channel_data [0,1] [A,B] [A,D] [3,4] [D,E] - - - - 2nd two_channel_data [2,3] [D,E] [B,E] - - - - - - three_channel_data - - - [0,1,2] [A,B,C] - - - - four_channel_data - - - - - [0,1,2,3] [A,B,D,E] - - five_channel_data - - - - - - - [0,1,2,3,4] [A,B,C,D,E]
The final output mapping can be derived using table 184.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)168
Table 184: Final channel mapping for 7_X_codec_mode = ASPX_ACPL_2
channel_mode Out 3/4/0.x � Ls
LrsRsRrs
� = �1 0
0 0
0 1
0 0
� × �DE�
�LR� = �A
B�
5/2/0.x � LLwR
Rw
� = �1 0
0 0
0 1
0 0
� × �AB�
�LsRs
� = �DE�
3/2/2.x � LVhlR
Vhr
� = �1 0
0 0
0 1
0 0
� × �AB�
�LsRs
� = �DE�
5.4 96 and 192 kHz decoding If Boolean b_hsf_ext is true, the ac4_hsf_ext_substream structure contains further spectral data, scalefactor data and spectral noise fill data extending the data beyond 24 kHz. If the decoder is not capable of decoding into 96 kHz or 192 kHz output, it shall ignore the additional coefficients.
If the decoder is capable of decoding into 96 kHz or 192 kHz output, it may be configured to these higher rates. In that case, it shall read the additional data up to either twice the original block length (for 96 kHz) or four times the original block length (for 192 kHz), and continue processing at twice or four times the block length and sampling rate.
NOTE: Streams containing high sampling frequency data do not need to employ any of the QMF domain tools.
5.5 IMDCT equations and block switching
5.5.1 Introduction
The choice of analysis block length is fundamental to any transform-based audio coding system. A long transform length is most suitable for input signals whose spectrum remains stationary, or varies only slowly, with time. A long transform length also provides greater frequency resolution, and hence improved coding performance for such signals. On the other hand, a shorter transform length, possessing greater time resolution, is more desirable for signals that change rapidly in time. Therefore, the time vs. frequency resolution trade-off should be considered when selecting a transform block length.
AC-4 makes this tradeoff by a flexible approach, providing full blocks and multiple partial blocks, which allows the codec to adapt the frequency/time resolution of the transform depending upon spectral and temporal characteristics of the signal being processed. The possible full block lengths, 2 048, 1 920, and 1 536, can subsequently be split into 2, 4, 8, or 16 sub-blocks.
Clause 5.5.2 and clause 5.5.3 contain information about the transform used and block-switching method in AC-4 decoders.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)169
5.5.2 Transforms
5.5.2.1 Data and control interfaces
The IMDCT applies the inverse of the frequency mapping that was carried out in the encoder. See clause 5.5.3 for a full list of supported transform lengths.
The input to the IMDCT is a block of N scaled spectral lines:
sIMDCT,ch Vector of N spectral lines of channel ch.
The output from the IMDCT and the overlap/add is a block of N time-domain reconstructed audio samples:
pIMDCT,ch Vector of N time-domain audio samples of channel ch.
The bitstream parameters used by the IMDCT and the overlap/add are:
N Block length, equal to the number of input spectral lines and to the number of output PCM samples. The window length is 2 × N.
Nfull Block length of a full block. This is equal to the frame_length as specified in table 83.
The state internal to the transforms is the overlap buffer:
Nprev Block length of previous block. This is set to N after finishing the processing of the current block.
overlap Vector of delayed time-domain audio samples. The delayed samples from the previous block, which have not been windowed, are used by the overlap/add step. Contains Nfull values from the processing of the previous block on input and provides Nfull values for the next block on output.
5.5.2.2 Decoding process
In the following procedure, steps 1 through 4 describe the computation of the IMDCT, which transforms the N spectral lines into an intermediate single real data block of length 2N using a single N/2 point complex IFFT with simple pre- and post-phase-shift operations. Steps 5 and 6 describe the unfolding, windowing and the overlap/add steps.
Step 1: IMDCT coefficient step.
Define the MDCT transform coefficients X[k] = sIMDCT,ch[k], k = 0, 1, …, N - 1.
Step 2: Pre-IFFT complex multiply step.
Compute N/2-point complex multiplication product Z[k]=Zr[k]+j Zi[k], k = 0, 1, …, N/2 - 1:
Pseudocode 60
for(k=0; k<N/2; k++) { /* Z[k] = (X[N-2*k-1] + j * X[2*k]) * (xcos1[k] + j * xsin1[k]); */ Zr[k] = (X[N-2*k-1] * xcos1[k] - X[2*k] * xsin1[k]); Zi[k] = (X[2*k] * xcos1[k] + X[N-2*k-1] * xsin1[k]); }
where:
xcos1[k] = -cos(2π x (8k + 1) / 16N),
xsin1[k] = -sin(2π x (8k + 1) / 16N),
and j is the imaginary unit.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)170
Step 3: Complex IFFT step.
Compute N/2-point complex IFFT of Z[k] to generate complex-valued sequence z[n] (using a direct formulation instead of the fast Fourier form for clarity):
Pseudocode 61
for(n=0; n<N/2; n++) { z[n] = 0; for(k=0; k<N/2; k++) { z[n] + = Z[k] * (cos(4*pi*k*n/N) + j * sin(4*pi*k*n/N)); } }
Step 4: Post-IFFT complex multiply step.
Compute N/2-point complex multiplication product y[n]=yr[n]+j yi[n], n = 0, 1, …, N/2 - 1 as:
Pseudocode 62
for(n=0; n<N/2; n++) { /* y[n] = z[n] * (xcos1[n] + j * xsin1[n]) / N; */ yr[n] = (zr[n] * xcos1[n] - zi[n] * xsin1[n]) / N; yi[n] = (zi[n] * xcos1[n] + zr[n] * xsin1[n]) / N; }
where:
zr[n] = real(z[n]),
zi[n] = imag(z[n]), and
xcos1[n] and xsin1[n] are as defined in step 2 above.
Step 5: Unfolding, left-half windowing and de-interleaving step.
Compute left-half windowed time-domain samples x[n]:
Pseudocode 63
for(n=0; n<N/4; n++) { x[2*n] = yi[N/4+n] * w[2*n]; x[2*n+1] = -yr[N/4-n-1] * w[2*n+1]; x[N/2+2*n] = yr[n] * w[N/2+2*n]; x[N/2+2*n+1] = -yi[N/2-n-1] * w[N/2+2*n+1]; x[N+2*n] = yr[N/4+n]; x[N+2*n+1] = -yi[N/4-n-1]; x[3*N/2+2*n] = -yi[n]; x[3*N/2+2*n+1] = yr[N/2-n-1]; }
where:
yr[n] = real(y[n]),
yi[n] = imag(y[n]), and
w[n] is the left transform window sequence which depends on Nprev and N:
w�n� = � 0, 0 ≤ � < Nskip
KBD_LEFT�NW,� − Nskip�, Nskip ≤ � < NW + Nskip
1, NW + Nskip ≤ � < NW + 2�skip
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)171
where:
NW = � N, N≤Nprev
Nprev, N>Nprev
Nskip = (� − �W)/2
and KBD_LEFT(N,n) is the left part of the Kaiser-Bessel derived window as specified in clause 5.5.3.
Step 6: Right-half windowing and overlap/add step.
The first half, i.e. the windowed part of the block, is overlapped with the second half of the previous block, after applying the windowing to the second half of the previous block, to produce PCM samples:
Pseudocode 64
nskip = (Nfull - N)/2; /* window second half of previous block */ nskip_prev = (Nfull - Nprev)/2; for (n=0; n<Nprev; n++) { overlap[nskip_prev+n] *= w[n]; } /* overlap/add using first N samples from x[n] */ for (n=0; n<N; n++) { overlap[nskip+n] += x[n]; } /* output pcm */ for (n=0; n<N; n++) { pcm[n] = overlap[n]; } /* move samples in overlap[] not stored in pcm[] */ for (n=0; n<nskip; n++) { overlap[n] = overlap[N+n]; } /* store second N samples from x[n] for next overlap/add */ for (n=0; n<N; n++) { overlap[nskip+n] = x[N+n]; }
where:
w[n] is the right transform window sequence which depends on Nprev and N:
w�n� = � 1, 0 ≤ � < Nskip
KBD_RIGHT�NW,� − Nskip�, Nskip ≤ � < NW + Nskip
0, NW + Nskip ≤ � < NW + 2�skip
where:
NW = � N, N≤Nprev
Nprev, N>Nprev
Nskip = (Nprev −�W)/2
and KBD_RIGHT(N,n) is the right part of the Kaiser-Bessel derived window as specified in clause 5.5.3.
The PCM output samples for each block pIMDCT,ch correspond to the content of the pseudocode array pcm.
The arithmetic processing in the overlap/add processing, if implemented in fixed-point arithmetic, shall use saturation arithmetic to prevent overflow (wrap-around). Since the output signal consists of the original signal plus coding error, it is possible for the output signal to exceed 100 % level even though the original input signal was less than or equal to 100 % level.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)172
5.5.3 Block switching
The following full and partial block lengths can be employed in the AC-4 transform block-switching procedure if the sampling frequency is 44,1 kHz.
• 2 048, 1 024, 512, 256, 128
If the sampling frequency is 48 kHz, the following additional block lengths are possible:
• 1 920, 960, 480, 240, 120
• 1 536, 768, 384, 192, 96
If the sampling frequency is 96 kHz or 192 kHz the values above shall be multiplied by 2 or 4, respectively.
KBD windows with the appropriate alpha values are used for the inverse MDCT operation. Table 185 shows the alpha values used for the KBD windows for the various transform lengths.
Table 185: Alpha values for KBD windows for different transform lengths
Sampling frequency α value for KBD window 44,1 kHz or 48 kHz 96 kHz 192 kHz
Transform sizes 2 048 1 920 1 536 4 096 3 840 3 072 8 192 7 680 6 144 3 1 024 960 768 2 048 1 920 1 536 4 096 3 840 3 072 4 512 480 384 1 024 960 768 2 048 1 920 1 536 4,5 256 240 192 512 480 384 1 024 960 768 5 128 120 96 256 240 192 512 480 384 6
The Kaiser-Bessel derived (KBD) windows are defined as follows:
KBD_LEFT(�,�) = �∑ W�N,p,α�np=0∑ W�N,p,α�Np=0
for 0≤n<N
KBD_RIGHT(N,n)=�∑ W�N,p,α�������p=0∑ W�N,p,α�N
p=0
for N≤n<2N
where:
W(N,n,α) is the Kaiser-Bessel kernel window function defined as:
W�N,n,α� = I�πα�1.0 − �2�N − 1 �!I(πα)
for 0≤n<N
where:
I�x� = "#�x2 k
k!$�∞
k=0
and α is the kernel window alpha factor as specified in table 185.
Larger block lengths have better frequency resolution but reduced time resolution. When the encoder detects a transient in the signal, it switches to shorter block lengths and indicates this in the coded bitstream. The size of the newly selected block length depends on the nature of the transient.
For an input block length of N, the number of intermediate windowed samples is 2N. When the current block and the preceding block are both of the same size, a constant window overlap of 50 % is used, and a history buffer stores the second half of the 2N samples, i.e. N samples. The first N samples of the current block are summed with windowed N samples of the previous block in the history buffer to produce the time-domain output samples.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)173
When a transient occurs, a switch to a specific smaller block length is signalled in the bitstream, and the corresponding window is then applied. To preserve the time-domain aliasing cancellation properties of the MDCT and IMDCT transforms and maintain block alignment, the window shape of the right half of the preceding block is modified when the following block has a smaller length. Because the window for the right half of each block can be applied when the overlap is done with the next block, the block length data is readily available by the time the overlap is calculated.
Similarly, when switching from a smaller block length to a larger block length, the window shape of the left half of the current block is suitably modified.
The example in figure 5 shows a full block length of N = 2 048 samples, and the switch to a partial block length of N/2 = 1 024 samples and subsequent switches to N/4 = 512 samples and back to N/2 = 1 024 samples. The change in window shape at block transitions can also be seen.
Figure 5: Block-switching during transient conditions
Partial block lengths are factors of 2, 4, 8, and 16 smaller than the full block lengths.
Two types of transitions are possible when switching from full blocks to partial blocks:
• switching from a full block to partial blocks of the same size, e.g. 2 048 to 2 x 1 024 or 2 048 to 4 x 512; and
• switching from a full block to a combination of partial block sizes, e.g. 2 048 to 1 x 1 024 + 8 x 128 or 2 048 to 2 x 512 + 1 x 1 024.
In either case, the sum total of all partial block sizes used is equal to the full block size.
Table 186 lists the various block-switching transitions permitted for initial block lengths of 2 048, 1 920, and 1 536 samples.
Table 186: Permitted block-switching transitions for full block lengths of 2 048, 1 920, and 1 536 samples
FULL BLOCK: 2 048 samples FULL BLOCK: 1 920 samples FULL BLOCK: 1 536 samples 1 024|1 024 960|960 768|768 1 024|2*512, 2*512|1 024 960|2*480, 2*480|960 768|2*384, 2*384|768 1 024|4*256, 4*256|1 024 960|4*240, 4*240|960 768|4*192, 4*192|768 1 024|8*128, 8*128|1 024 960|8*120, 8*120|960 768|8*96, 8*96|768 2*512|2*512 2*480|2*480 2*384|2*384 2*512|4*256, 4*256|2*512 2*480|4*240, 4*240|2*480 2*384|4*192, 4*192|2*384 2*512|8*128, 8*128|2*512 2*480|8*120, 8*120|2*480 2*384|8*96, 8*96|2*384 4*256|4*256 4*240|4*240 4*192|4*192 4*256|8*128, 8*128|4*256 4*240|8*120, 8*120|4*240 4*192|8*96, 8*96|4*192 8*256 8*240 8*192 16*128 16*120 16*96
EXAMPLE: One mono channel with frame_length = 1 920, consisting of 2 partial blocks of size 480 and 1 partial block of size 960 (transf_length[0] = 2 and transf_length[1] = 3). The previous frame contained a full block:
� The first 480 spectral lines are inverse MDCT transformed. Then, the first 480 of the resulting 960 time samples are windowed using an unmodified KBD left 480 window since the preceding block did not have a shorter length. The samples in the overlap add buffer are windowed using an extended KBD right 480 window of size 1 920: the first 720 samples are unmodified, the next 480 samples are windowed using the KBD right 480 window and the last 720 samples are set to 0. Next, the 480 left windowed samples are added on a sample-by-sample basis using a factor of 2 to the 480 samples in the overlap add buffer starting at offset 720. The first 480 samples of the overlap add buffer are passed on to the PCM output buffer.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)174
� The next 720 samples in the overlap add buffer are moved to the beginning of the buffer and the second 480 IMDCT output samples, the unwindowed samples, are stored in the overlap add buffer starting at offset 720.
� The second 480 spectral lines are inverse MDCT transformed. Then, the first 480 of the resulting 960 time samples are windowed using an unmodified KBD left 480 window since the preceding block did not have a shorter length. The samples in the overlap add buffer are windowed using a KBD right 480 window starting at sample 720. Next, the 480 left windowed samples are added on a sample-by-sample basis using a factor of 2 to the 480 samples in the overlap add buffer starting at offset 720. The first 480 samples of the overlap add buffer are passed on to the PCM output buffer. The next 720 samples in the overlap add buffer are moved to the beginning of the buffer and the second 480 IMDCT output samples, the unwindowed samples, are stored in the overlap add buffer starting at offset 720.
� The third block of 960 spectral lines is inverse MDCT transformed. Then, the first 960 of the resulting 1 920 time samples are windowed using an extended KBD left 480 window of size 960 since the preceding block has a shorter length: the first 240 samples are set to 0, the next 480 samples are windowed using the KBD left 480 window and the following 240 samples are left unmodified. Next, the 960 left windowed samples are added on a sample-by-sample basis using a factor of 2 to the 960 samples in the overlap add buffer starting at offset 480. The first 960 samples of the overlap add buffer are passed on to the PCM output buffer. The next 480 samples in the overlap add buffer are moved to the beginning of the buffer and the second 960 IMDCT output samples, the unwindowed samples, are stored in the overlap add buffer starting at offset 480.
In this manner, all the blocks of the current frame are processed into a composition buffer. The composition buffer needs to hold at most 4 096 samples (8 192 for a 96 kHz decoder or 16 384 for a 192 kHz decoder). After processing all blocks of the current frame, the earliest frame_length samples are passed on to the next tool in the processing chain, and removed from the composition buffer. Silent samples are filled in on the right.
5.6 Frame alignment
5.6.1 Introduction
AC-4 is designed to enable artefact-free switching and splicing between different streams, as can happen in adaptive streaming [1]. When the input of the codec is switched to a different adaptation set, it is important to take into account the codec internal signal and metadata delays.
Because some control data apply to entire codec frames and cannot easily be delayed by fractional amounts, the PCM domain signal is delayed to make the control data delays multiples of integer frames.
Data and control interfaces
The input to the frame alignment tool is the time-domain signal output from the IMDCT:
pinFA,ch Vector of frame_length time-domain audio samples of channel ch.
The output from the frame alignment tool are delayed time-domain audio samples:
poutFA,ch Vector of frame_length time-domain audio samples of channel ch.
5.6.2 Decoding process
A decoder shall apply a delay between the input and output PCM samples according to:
out�n� = %�[� − &_'()]
where d_pcm is defined per table 187.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)175
Table 187: Delay in frame alignment tool
frame_rate PCM signal delay d_pcm [samples] Control data delay d_ctrl [frames] 23,976 288 1 24 288 1 25 352 1 29,97 96 1 30 96 1 47,952 960 2 48 960 2 50 1 056 2 59,94 672 2 60 672 2 100 1 312 4 119,88 864 4 120 864 4 21,5332 (see note 1) 352 1 23,4375 (see note 2) 352 1 NOTE 1: Music at 44,1 kHz. NOTE 2: Music at 48,0 kHz.
5.7 QMF domain processing
5.7.1 Introduction
AC-4 decoders employ a complex QMF analysis/synthesis filter bank pair to enable alias suppressed frequency-domain processing. The output of the spectral frontend is transformed into a matrix of num_qmf_subbands complex QMF subbands as rows and num_qmf_timeslots time slots as columns, where num_qmf_timeslots is equal to (frame_length/num_qmf_subbands). The entire output of a block from the spectral frontend is computed before QMF domain processing commences.
To enable tools to transition smoothly across blocks, 6 QMF time slots of history are kept in between blocks. This means that the QMF synthesis works on QMF data that is delayed by 6 QMF time slots, or 6 × num_qmf_subbands time domain samples.
5.7.2 QMF control data alignment
The control data of tools operating in QMF domain is at all times aligned with the corresponding spectral frontend data contained in the sf_data element. This means that within a raw AC-4 frame, the control data of any QMF domain tool shall be applied to the spectral data encoded in the very same raw AC-4 frame, regardless of the delay introduced by transforms or other tools. This is done to allow for a smoother switching behaviour when decoding spliced AC-4 bitstreams.
For the decoder, this property means that it shall:
1) Decode the control data of the QMF domain tool.
2) Keep this control data on hold to take care of the delay introduced in the signal chain.
3) Apply it as soon as the corresponding signal data is processed by the respective QMF domain tool.
The delay introduced by the signal chain is different for different values of frame_length. However, due to the delay introduced in the frame alignment in clause 5.6, it always corresponds to an integer number of AC-4 frames. See value d_ctrl in table 187.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)176
5.7.3 QMF analysis filterbank
5.7.3.1 Data and control interfaces
A QMF bank is used to split the time domain signal from the core decoder into num_qmf_subbands subband signals. The output from the filterbank, i.e. the subband samples, is complex valued and thus oversampled by a factor two compared to a regular cosine modulated QMF bank.
The input of the QMF analysis filterbank are frame-aligned time domain audio samples.
pinQMF,ch Vector of frame_length time domain audio samples of channel ch.
The output of the QMF analysis filterbank is a QMF matrix.
QoutQMF,ch Matrix of num_qmf_subbands rows and num_qmf_timeslots columns of channel ch.
Each column of QoutQMF,ch is a vector of num_qmf_subbands complex values, one value for each QMF subband, representing a QMF time slot. The rows of QoutQMF,ch represent the QMF subband signals.
The bitstream and additional parameters used by the QMF analysis filterbank are:
frame_length Frame length, equals the number of input PCM samples.
num_qmf_timeslots
Number of QMF time slots, derived from frame_length. num_qmf_win_coef
Number of coefficients in the QMF window, 640. QWIN[num_qmf_win_coef]
QMF window coefficients, see table D.3.
The filter state is kept in:
qmf_filt Vector of 640 delayed time-domain audio samples in reversed order, where 576 samples are saved in between calls.
5.7.3.2 Decoding process
The number of QMF subbands in the analysis QMF filter bank, num_qmf_subbands, is always 64. The number of QMF slots is calculated as follows:
num_qmf_timeslots=frame_length
num_qmf_subbands
Table 188 shows the possible values for num_qmf_timeslots.
Table 188: Number of QMF time slots
frame_length 2 048 1 920 1 536 1 024 960 768 512 384 num_qmf_timeslots 32 30 24 16 15 12 8 6
The filtering involves the following steps, which are also represented in the pseudocode that follows. A higher index into the vector corresponds to older samples.
• Shift the samples in the vector qmf_filt by num_qmf_subbands positions. The oldest num_qmf_subbands samples are discarded and num_qmf_subbands new samples are stored in positions 0 to num_qmf_subbands - 1.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)177
• Multiply the samples of vector qmf_filt by window coefficients from QWIN to produce vector z. The window coefficients can be found in table D.3.
• Sum the samples according to the formula in the pseudocode to create the 2 × num_qmf_subbands-element vector u.
• Calculate 64 new subband samples by the matrix operation M × u, where:
Mk,n = exp � j·π·�k+0.5� ∙ �2� − 1�2 ∙ �+)_,)-_.+//0�&.! , �0 ≤ 1 < �+)_,)-_.+//0�&.
0 ≤n<2 · num_qmf_subbands
In the equation, exp() denotes the complex exponential function and j is the imaginary unit.
Every loop represented in the following pseudocode produces num_qmf_subbands complex-valued subband samples, each representing the output from one filterbank subband. In the pseudocode Q[sb][ts] corresponds to the QMF subband sample sb of time slot ts.
Pseudocode 65
for (ts = 0; ts < num_qmf_timeslots; ts++) { /* shift time-domain input samples by 64 */ for (sb = 639; sb >= 64; sb--) { qmf_filt[sb] = qmf_filt[sb-64]; } /* feed new audio samples */ for (sb = 64-1; sb >= 0; sb--) { qmf_filt[sb] = pcm[ts*64+63-sb]; } /* multiply input samples by window coefficients */ for (n = 0; n < 640; n++) { z[n] = qmf_filt[n] * QWIN[n]; } /* sum the samples to create vector u */ for (n = 0; n < 128; n++) { u[n] = z[n]; for (k = 1; k < 5; k++) { u[n] = u[n] + z[n + k*128]; } } /* compute 64 new subband samples */ for (sb = 0; sb < 64; sb++) { /* note that Q[sb][ts] is a complex datatype */ Q[sb][ts] = u[0] * exp(j*(pi/128)*(sb+0.5)*(-1)); for (n = 1; n < 128; n++) { Q[sb][ts] += u[n] * exp(j*(pi/128)*(sb+0.5)*(2*n - 1); } } }
5.7.4 QMF synthesis filterbank
5.7.4.1 Data and control interfaces
Synthesis filtering of the QMF subband signals is achieved using a num_qmf_subbands-subband synthesis QMF bank. The output from the filterbank is real-valued time domain samples.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)178
The input to the QMF synthesis filterbank is a set of num_qmf_subbands complex-valued subband signals:
QinQMF,ch Matrix of num_qmf_subbands rows and num_qmf_timeslots columns of channel ch.
Each column of QinQMF,ch is a vector of num_qmf_subbands complex values, one value for each QMF subband, representing a QMF time slot. The rows of QinQMF,ch represent the QMF subband signals.
The output of the QMF synthesis filterbank are time domain audio samples:
poutQMF,ch Vector of frame_length time domain audio samples of channel ch.
The bitstream and additional parameters used by the QMF synthesis filterbank are:
frame_length Frame length, equals the number of output PCM samples.
num_qmf_timeslots Number of QMF slots, derived from frame_length.
num_qmf_win_coef Number of coefficients in the QMF window, 640.
QWIN[num_qmf_win_coef] QMF window coefficients, see table D.3.
The filter state is kept in:
qsyn_filt Vector of 1 280 synthesis time domain samples, where 1 152 samples are saved in between calls.
5.7.4.2 Decoding process
The synthesis filtering comprises the following steps, where a vector qsyn_filt consisting of 1 280 samples is assumed:
1) Shift the samples in the vector qsyn_filt by 128 positions. The oldest 128 samples are discarded.
2) The num_qmf_subbands new complex-valued subband samples are multiplied by the matrix N , where:
Nn,k = 1
num_qmf_subbands
∙ exp �j·π·�k+0.5� ∙ �2� − 4*num_qmf_subbands-1�2 ∙ �+)_,)-_.+//0�&. ! , �0 ≤ 1 < �+)_,)-_.+//0�&.
0 ≤n<2·num_qmf_subbands
In the equation, exp() denotes the complex exponential function and j is the imaginary unit. The real part of the output from this operation is stored in the positions 0 to 2*num_qmf_subbands-1 of vector qsyn_filt.
3) Extract samples from qsyn_filt to create the 10×num_qmf_subbands-element vector g.
4) Multiply the samples of vector g by window QWIN to produce vector w. The window coefficients QWIN can be found in table D.3, and are the same as for the analysis filterbank.
5) Calculate num_qmf_subbands new output samples by summation of samples from vector w.
In the following pseudocode, Q [sb][ts] corresponds to the QMF timeslot ts in the QMF subband sb, and every new loop produces num_qmf_subbands time-domain samples as output.
Pseudocode 66
for (ts = 0; ts < num_qmf_timeslots; ts++) { /* shift samples by 128 */ for (n = 1279; n >= 128; n--) { qsyn_filt[n] = qsyn_filt[n-128]; } for (n = 0; n < 128; n++) {
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)179
exponent = j*(pi/128)*(0.5)*(2*n - 255)); qsyn_filt[n] = real(Q[0][ts]/64 * exp(exponent)); for (sb = 1; sb < num_qmf_subbands; sb++) { exponent = j*(pi/128)*(sb+0.5)*(2*n - 255); qsyn_filt[n] += real(Q[sb][ts]/64 * exp(exponent)); } } for (n = 0; n < 5; n++) { for (sb = 0; sb < 64; sb++) { g[128*n + sb] = qsyn_filt[256*n + sb]; g[128*n + 64 + sb] = qsyn_filt[256*n + 192 + sb]; } } /* multiply by window coefficients */ for (n = 0; n < 640; n++) { w[n] = g[n] * QWIN[n]; } /* compute 64 new time-domain output samples */ for (sb = 0; sb < 64; sb++) { temp = w[sb]; for (n = 1; n < 10; n++) { temp = temp + w[64*n + sb]; } pcm[ts*64 + sb] = temp; } }
5.7.5 Companding tool
5.7.5.1 Data and control interfaces
The companding tool is used to mitigate pre- and post-echo artefacts. It is applied on the QMF domain data. The encoder applies attenuation to signal parts with higher energy, and applies gain to signal parts with lower energy. The decoder applies the inverse process, which effectively shapes the coding noise by the signal energy.
The input to and output from the companding tool is M QMF matrices, with M being the number of channels obtained from the channel element of the AC-4 frame. The matrices comprise exactly those QMF time slots that are part of the A-SPX interval, described in clause 5.7.6.3.3.1.
Input
QinCOMP,[a|b|…] M complex QMF matrices of the M channels to be processed by the companding tool.
Output
QoutCOMP,[a|b|…] M complex QMF matrices of the same M channels processed by the companding tool.
The elements of the matrices QinCOMP,ch and QoutCOMP,ch shall be denoted as Qinch(sb,ts) and Qoutch(sb,ts), respectively, for each QMF subband sb and each QMF time slot ts.
The bitstream and additional information used by the companding tool is:
b_compand_onch See clause 4.3.9.
b_compand_avg
See clause 4.3.9. sync_flag
See clause 4.3.9.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)180
acpl_qmf_bandch Lower subband border, needed if the codec mode is ASPX_ACPL_1.
aspx_xover_bandch
A-SPX crossover subband for channel ch.
NOTE: The variable aspx_xover_band is derived as sbx in clause 5.7.6.3.1.2.
5.7.5.2 Decoding process
The QMF subband range [sb0, sb1-1] that the companding tool works on shall be determined according to:
sb� = �acpl_qmf_bandch in case the codec mode is ASPX_ACPL_1
0 otherwise
sb� = aspx_xover_bandch
The QMF timeslot range [ts0, ts1-1] that the companding tool works on shall be determined according to:
ts� = atsg_sig[0] × num_ts_in_ats and ts� = atsg_sig[num_atsg_sig] × num_ts_in_ats
Let K=sb1-sb0 be the number of affected subbands.
Absolute sample levels Ech(sb,ts) shall be calculated according to:
Ech�sb,ts� = )02�3Re4Qinch�sb,ts�53, 3Im4Qinch�sb,ts�53� + 0,5 ∗ min�3Re4Qinch(./, 6.)53, 3Im4Qinch(./, 6.)53�
A slot mean absolute level Lch(ts) shall be calculated according to:
Lch�ts� = 0,9105 ×1
K7 Ech�sb, ts�sb���
sbsb�
The per-slot gain gch(ts) shall be computed as:
gch(6.) = Lch�ts�(��) α ⁄ where α = 0,65.
Processing if sync_flag = 0:
If b_compand_onch is TRUE, the gain gch(ts) shall be applied to the affected subbands for each time slot in the A-SPX interval:
Qoutch(6., ./) = gch(6.) × 9 × Qinch(6., ./) where G=0,5�� α⁄ .
If b_compand_onch is FALSE and b_compand_avg is TRUE, Lch(ts) shall be averaged over the entire A-SPX interval:
Lavg, ch =1
num_qmf_timeslots_in_aspx_int7 Lch�ts�ts�
tsts�
and the average gain shall be calculated as:
gavg,ch = Lavg,ch
�1-:�
and applied as:
Qoutch(6., ./) = gavg,ch × 9 × Qinch(6., ./) Processing if sync_flag = 1:
If all channels are synched (sync_flag = 1), then the synched per-slot gain gsynch(ts) is computed using the per-slot gain gch(ts) as:
gsynch�ts� = 1
M7 gch
�ts�M-1
ch=0
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)181
If b_compand_on0 is TRUE, the gain gsynch(ts) shall be applied to the affected subbands for each time slot in the A-SPX interval and for each of the M channels:
Qoutch(6., ./) = gsynch�ts� × 9 × Qinch(6., ./)
If b_compand_on0 is FALSE and b_compand_avg is TRUE, gsynch(ts) shall be averaged over the entire A-SPX interval:
gavg, synch =1
num_qmf_timeslots_in_aspx_int7 gsynch(6.)ts1
ts=ts0
The average synched gain shall be applied as:
Qoutch(6., ./) = gavg,synch × 9 × Qinch(6., ./) 5.7.6 Advanced spectral extension tool - A-SPX
5.7.6.1 Data and control interfaces
The A-SPX tool is used to parametrically code the higher frequencies of the audio signal.
IWC (interleaved waveform coding) can be employed in addition, if the parametric model cannot re-instate the correct frequency components, or to achieve a more accurate temporal reproduction of the signal.
Input:
QinASPX,a QMF subsamples for channel a.
QinASPX,b Second complex QMF matrix containing the QMF subsamples for an optional second channel b.
Output:
QoutASPX,a Complex QMF matrix containing the A-SPX processed QMF subsamples for channel a.
QoutASPX,b Second complex QMF matrix containing the A-SPX processed QMF subsamples for channel b, in the case of two input channels.
The QinASPX and QoutASPX matrices each consist of num_qmf_timeslots columns, and num_qmf_subband rows.
Control:
• Dequantized control data, which are inputs to the HF generator.
• Huffman decoded and dequantized A-SPX signal and noise envelope data, which are inputs to the HF envelope adjuster.
• Signalling information used by the interleaved waveform coding block.
Figure 6 illustrates the key blocks of the A-SPX tool as well as its position in the processing chain.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)182
Figure 6: A-SPX block diagram
The output of the IMDCT for channel a is frame aligned and routed into the analysis QMF bank. The filterbank generates a complex QMF matrix containing the QMF subsamples. This matrix is subsequently processed by the companding tool before being fed into the A-SPX tool as QinASPX,a.
If channel a is one of two channels of an aspx_data_2ch element, the companion channel b is processed the same way as channel a, resulting in QinASPX,b. If not stated differently, both channels are processed likewise.
In the A-SPX tool, the matrix QinASPX is low band filtered at the cross over frequency to yield the matrix QLow.
Using the parameter information transmitted in the AC-4 bitstream, and given the matrix QLow, the HF generator performs subband tonal to noise ratio adjustment and patching operations to derive an estimate of the high-frequency components of the spectrum, the matrix QHigh.
The envelope adjuster then corrects the estimated spectrum QHigh by using decoded A-SPX signal and noise envelope data from the bitstream and produces matrix Y.
The interleaved waveform coding block takes a delayed matrix QinASPX and the matrix Y as inputs and produces an output matrix QoutASPX, which contains the low band, spectral extension components and interleaved waveform coded components.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)183
The matrix QoutASPX is finally routed into the synthesis QMF bank unless additional processing occurs in the QMF domain.
5.7.6.2 A-SPX specific variables
sbg_master Array of master QMF subband groups, of length num_sbg_master + 1.
sbg_sig_highres Array of QMF subband group borders for high frequency resolution signal envelopes; the array is of length num_sbg_sig_highres + 1.
sbg_sig_lowres Array of QMF subband group borders for low frequency resolution signal envelopes; the array is of length num_sbg_sig_lowres + 1.
sbg_lim Array is of length num_sbg_lim + 1 and contains frequency borders used by the limiter.
sbg_noise Array of QMF subband group borders for adaptive noise envelope addition; the array is of length num_sbg_noise + 1.
sbg_patches Array of QMF subband group borders for HF patches; the array is of length num_sbg_patches + 1.
sbg_sig Matrix of QMF subband group borders for low-frequency and high-frequency resolution signal envelopes; it contains 2 columns, column zero is of length num_sbg_sig_lowres + 1, and column one is of length num_sbg_sig_highres + 1.
scf_sig_sbg Signal envelope scale factors; matrix of num_atsg_sig columns with each column env containing either num_sbg_sig[0] or num_sbg_sig[1] rows.
scf_noise_sbg Noise envelope scale factors; matrix of num_atsg_noise columns and num_sbg_noise rows.
sbx First QMF subband in the A-SPX range.
sba First QMF subband in the sbg_master table.
num_sb_aspx Number of QMF subbands in the A-SPX range.
num_sbg_lim Number of limiter subband groups.
num_sbg_master Number of subband groups in the master subband group array.
num_sbg_noise Number of noise envelope subband groups.
num_sbg_sig Array with two entries for the number of subband groups for signal envelopes: entry 0 for low resolution envelopes and entry 1 for high resolution envelopes.
num_sbg_patches Number of patch subband groups.
num_aspx_timeslots Number of A-SPX time slots.
atsg_sig Array containing start and stop time borders for all signal envelopes in the current A-SPX interval; the array is of length num_atsg_sig + 1.
atsg_noise Array containing start and stop time borders for all noise envelopes in the current A-SPX interval; the array is of length num_atsg_noise + 1.
num_atsg_sig Number of signal envelopes in an A-SPX interval.
num_atsg_noise Number of noise envelopes in an A-SPX interval.
ts_offset_hfadj Time slot offset for the HF envelope adjuster.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)184
ts_offset_hfgen Time slot offset for the HF generator.
Qlow, Q_low[sb][ts] Complex input QMF subband matrix to the HF generator; it contains num_qmf_timeslots columns and num_qmf_subbands rows.
Qhigh, Q_high[sb][ts] Complex output QMF subband matrix of the HF generator; it contains num_qmf_timeslots columns and num_sb_aspx rows.
Y Complex output QMF bank subband matrix from the HF envelope adjuster.
5.7.6.3 Decoding A-SPX control data
5.7.6.3.1 Subband groups
5.7.6.3.1.0 Introduction
QMF subband samples are combined to form subband groups (sbg). These are described by subband group tables of the following types:
Template subband group tables, sbg_template_[high|low]res Static high-resolution and low-resolution subband group tables serving as a template for deriving the master table.
Master subband group table for signal envelopes, sbg_master
This is the table from which all the other subband group tables are either directly derived (high-resolution and low-resolution subband group tables) or indirectly derived (noise and limiter subband group tables).
Subband group table for high/low-resolution signal envelopes sbg_sig_[high|low]res
This table contains the subband borders of signal envelopes with high/low frequency resolution. Subband group table for noise envelopes, sbg_noise
This table contains the subband borders of noise envelopes. Subband group table for the limiter, sbg_lim
This table is contains subband borders used by the limiter. Subband group table for patches, sbg_patches
This table is contains subband borders used by HF patch generator.
The subband group tables contain the QMF subbands that mark the lower borders of each group. Consecutive groups lie side by side in the spectrum; hence, the upper borders of the groups are implicitly defined by the lower border of the next higher group. Each subband group includes its lower border QMF subband, but not the higher border one. Additionally, the subband group tables contain the upper border of the highest subband group.
The number of subband groups in each table is described by the notation num_[subband_group_name], e.g. num_sbg_sig_highres. The number of subband borders in each table is therefore num_[subband_group_name] + 1.
The parameters required by the decoder to calculate the subband group tables are transmitted in the A-SPX header, described by the bitstream element aspx_config. The headers are only sent with every I-frame. Hence, the parameters sent in each aspx_config are valid until new parameter information is transmitted in a subsequent I-frame.
If the decoder does not have this header information, e.g. in the case of an irregularly spliced bitstream where the subsequent AC-4 frame is not at an I-frame, it is unable to replicate high frequencies until a valid A-SPX header is received.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)185
5.7.6.3.1.1 Master subband group table
Two static subband group tables are defined: one for low-frequency resolutions (sbg_template_lowres) and one for high-frequency resolutions (sbg_template_highres). The master table is derived from these statically pre-defined tables.
The template subband group tables are defined as:
sbg_template_lowres= �10,11,12,13,14,15,16,17,18,19,20,22,24,26,28,30,32,35,38,42,46� sbg_template_highres= �18,19,20,21,22,23,24,26,28,30,32,34,36,38,40,42,44,47,50,53,56,59,62�
The high-frequency resolution table supports up to 22 subband groups ranging from QMF subband 18 to subband 62. The low-frequency resolution table starts at QMF subband 10 and goes to subband 46, having up to 20 subband groups.
The following three parameters are needed to derive the current master table from the static template tables:
1) A-SPX start frequency (aspx_start_freq): a 3-bit (0-7) index into the template tables starting from the first QMF subband (18 or 10) and moving upwards with steps of 2. An aspx_start_freq of 1 hence points to QMF subband 20 in the high resolution template table.
2) A-SPX stop frequency (aspx_stop_freq): a 2-bit (0-3) index into the template tables starting from the last QMF subband (62 or 46) and going downwards with steps of 2. An aspx_stop_freq of 2 hence points to QMF subband 50 in the high resolution template table.
3) A-SPX master frequency table scale (aspx_master_freq_scale): a 1-bit (0-1) value indicating which of the two static template tables is currently used. aspx_master_freq_scale = 0 is the low-resolution template table and aspx_master_freq_scale = 1 indicates its high-resolution counterpart.
Table 189 and table 190 list the possible start and stop subband for the two static template tables, and the corresponding frequencies in Hz with respect to a 48 kHz sampling frequency.
Table 189: Master table start/stop subbands for low-frequency resolution template table
Low resolution subband group table (aspx_master_freq_scale = 0) aspx_start_freq QMF subband Frequency [Hz] aspx_stop_freq QMF subband Frequency [Hz] 0 10 3 750 0 46 17 250 1 12 4 500 1 38 14 250 2 14 5 250 2 32 12 000 3 16 6 000 3 28 10 500 4 18 6 750 - - - 5 20 7 500 - - - 6 24 9 000 - - - 7 28 10 500 - - -
Table 190: Master table start/stop subbands for high-frequency resolution template table
High-resolution subband group table (aspx_master_freq_scale = 1) aspx_start_freq QMF subband Frequency [Hz] aspx_stop_freq QMF subband Frequency [Hz] 0 18 6 750 0 62 23 250 1 20 7 500 1 56 21 000 2 22 8 250 2 50 18 750 3 24 9 000 3 44 16 500 4 28 10 500 - - - 5 32 12 000 - - - 6 36 13 500 - - - 7 40 15 000 - - -
The master subband group table, sbg_master, the number of master subband groups, num_sbg_master, the lower border of the first subband group, sba, and the upper border of the last subband group, sbz, are derived from the following pseudocode.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)186
Pseudocode 67
if (master_reset == 1) { if (aspx_master_freq_scale == 1) { num_sbg_master = 22 - 2 * aspx_start_freq - 2 * aspx_stop_freq; for (sbg = 0; sbg <= num_sbg_master; sbg++) { sbg_master[sbg] = sbg_template_highres[2 * aspx_start_freq + sbg]; } } else { num_sbg_master = 20 - 2 * aspx_start_freq - 2 * aspx_stop_freq; for (sbg = 0; sbg <= num_sbg_master; sbg++) { sbg_master[sbg] = sbg_template_lowres[2 * aspx_start_freq + sbg]; } } } sba = sbg_master[0]; sbz = sbg_master[num_sbg_master];
The master_reset variable is set to 1 if any of the following parameters has changed compared to the values sent in the previous I-frame:
aspx_master_freq_scale
aspx_start_freq
aspx_stop_freq
5.7.6.3.1.2 Signal envelope subband group tables
The subband group table for high-frequency resolution signal envelopes, sbg_sig_highres, covers the upper subband groups from sbg_master. For that, sbg_master gets split into two parts using the A-SPX cross over subband offset, aspx_xover_subband_offset. The aspx_xover_subband_offset element is an index into the master subband group table, starting at the first band and moving upward with a step of 1 subband group.
Hence, sbg_sig_highres is a truncated version of sbg_master, starting at the cross over subband, as shown in the following pseudocode.
Pseudocode 68
num_sbg_sig_highres = num_sbg_master - aspx_xover_subband_offset; for (sbg = 0; sbg <= num_sbg_sig_highres; sbg++) sbg_sig_highres[sbg] = sbg_master[sbg + aspx_xover_subband_offset]; sbx = sbg_sig_highres[0]; num_sb_aspx = sbg_sig_highres[num_sbg_sig_highres] - sbx;
The additional variables derived here are the number of subbands in the A-SPX range, num_sb_aspx and the cross over subband, sbx.
The low-resolution subband group table for signal envelopes, sbg_sig_lowres, is always a perfect decimation of its high- resolution counterpart by a factor 2. It is derived from sbg_sig_highres together with the number of subbands for low-resolution signal envelopes, num_sbg_sig_lowres, as shown in the following pseudocode.
Pseudocode 69
num_sbg_sig_lowres = num_sbg_sig_highres - floor(num_sbg_sig_highres/2); sbg_sig_lowres[0] = sbg_sig_highres[0]; if (mod(num_sbg_sig_highres, 2) == 0) { /* num_sbg_sig_highres even */ for (sbg = 1; sbg <= num_sbg_sig_lowres; sbg++) sbg_sig_lowres[sbg] = sbg_sig_highres[2*sbg]; } else { /* num_sbg_sig_highres odd */ for (sbg = 1; sbg <= num_sbg_sig_lowres; sbg++) sbg_sig_lowres[sbg] = sbg_sig_highres[2*sbg-1]; }
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)187
num_sbg_sig[0] = num_sbg_sig_lowres; num_sbg_sig[1] = num_sbg_sig_highres;
An additional array, num_sbg_sig, containing both the numbers of the high-resolution and low-resolution signal envelope subband groups, is derived here for convenience.
As seen from table 189 and table 190, the subband group tables always start and end on an even-numbered QMF subband. For some choices of aspx_xover_subband_offset, the first subband group of the high-resolution and low-resolution subband group table are identical.
5.7.6.3.1.3 Noise subband group table
The noise envelope subband group table, sbg_sig_noise, is obtained from sbg_sig_lowres as shown in the following pseudocode.
Pseudocode 70
num_sbg_noise = max(1, floor(aspx_noise_sbg * log2(sbz/sbx) + 0.5)); idx[0] = 0; sbg_noise[0] = sbg_sig_lowres[0]; for (sbg = 1; sbg <= num_sbg_noise; sbg++) { idx[sbg] = idx[sbg-1]; idx[sbg] += floor((num_sbg_sig_lowres - idx[sbg-1])/(num_sbg_noise + 1 - sbg)); sbg_noise[sbg] = sbg_sig_lowres[idx[sbg]]; }
The number of noise envelope subband groups, num_sbg_noise, shall satisfy num_sbg_noise ≤ 5.
5.7.6.3.1.4 Patch subband group table
The patch subband group tables, sbg_patch_num_sb, sbg_patch_start_sb, and the number of patches, num_sbg_patches, as well as the derived table containing the patch borders in the A-SPX range, sbg_patches, are created using the following pseudocode.
Pseudocode 71
msb = sba; usb = sbx; num_sbg_patches = 0; if (base_samp_freq == 48) goal_sb = 43; else goal_sb = 46; if (aspx_master_freq_scale == 1) source_band_low = 4; else source_band_low = 2; if (goal_sb < sbx + num_sb_aspx) { for (i = 0, sbg = 0 ; sbg_master[i] < goal_sb; i++) { sbg = i + 1; } } else { sbg = num_sbg_master; } do { j = sbg; sb = sbg_master[j]; odd = (sb - 2 + sba) % 2; while (sb > ( sba - source_band_low + msb - odd )) { j--; sb = sbg_master[j]; odd = (sb - 2 + sba) % 2; } sbg_patch_num_sb[num_sbg_patches] = max(sb - usb, 0); sbg_patch_start_sb[num_sbg_patches] = sba – odd – max(sb - usb, 0); if (sbg_patch_num_sb[num_sbg_patches] > 0) {
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)188
usb = sb; msb = sb; num_sbg_patches = num_sbg_patches + 1; } else { msb = sbx; } if (sbg_master[sbg] - sb < 3) { sbg = num_sbg_master; } } while (sb != (sbx + num_sb_aspx)); if ((sbg_patch_num_sb[num_sbg_patches-1] < 3) && (num_sbg_patches > 1)) { num_sbg_patches--; } sbg_patches[0] = sbx; for (i = 1; i <= num_sbg_patches; i++) { sbg_patches[i] = sbg_patches[i-1] + sbg_patch_num_sb[i-1]; }
The number of patches, num_sbg_patches, shall satisfy num_sbg_patches≤5.
Patch table generation is also influenced by the base sampling frequency, base_samp_freq.
5.7.6.3.1.5 Limiter subband group table
The limiter subband group table, sbg_lim, and the number of limiter subband groups, num_sbg_lim, is derived from the low frequency resolution table, sbg_sig_lowres, and the patch table, sbg_patches, as shown in the following pseudocode. It has two limiter subband groups per octave.
Pseudocode 72
/* Copy sbg_sig_lowres into lower part of limiter table */ for (sbg = 0; sbg <= num_sbg_sig_lowres; sbg++) { sbg_lim[sbg] = sbg_sig_lowres[sbg]; } /* Copy patch borders into higher part of limiter table */ for (sbg = 1; sbg < num_sbg_patches; sbg++); { sbg_lim[sbg+num_sbg_sig_lowres] = sbg_patches[sbg]; } /* Sort patch borders + low res sbg into temporary limiter table */ sort(sbg_lim); sbg = 1; num_sbg_lim = num_sbg_sig_lowres + num_sbg_patches - 1; while (sbg <= num_sbg_lim) { num_octaves = log2(sbg_lim[sbg] / sbg_lim[sbg-1]); if (num_octaves < 0.245) { if (sbg_lim[sbg] == sbg_lim[sbg-1]) { sbg_lim = remove_element(sbg_lim, num_sbg_lim, sbg); num_sbg_lim--; continue; } else { if (is_element_of_sbg_patches(sbg_lim[sbg])) { if (is_element_of_sbg_patches(sbg_lim[sbg-1])) { sbg++; continue; } else { sbg_lim = remove_element(sbg_lim, num_sbg_lim, sbg-1); num_sbg_lim--; continue; } } else { sbg_lim = remove_element(sbg_lim, num_sbg_lim, sbg); num_sbg_lim--; continue; } } } else { sbg++; continue;
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)189
} }
The is_element_of_sbg_patches() and remove_element() functions are defined as follows.
Pseudocode 73
is_element_of_sbg_patches(sbg_lim[sbg]) { for (i = 0; i <= num_sbg_patches; i++) { if (sbg_patches[i] == sbg_lim[sbg]) return TRUE; } return false; }
and
Pseudocode 74
remove_element(sbg_lim, num_sbg_lim, sbg) { for (i = sbg; i < num_sbg_lim; i++) { sbg_lim[i] = sbg_lim[i+1]; } return sbg_lim; }
5.7.6.3.2 Low band filter and QMF delay line
The QMF time-frequency samples from the analysis filterbank, stored in QinASPX, are truncated at the crossover subband, sbx, and delayed by ts_offset_hfgen before being fed into the HF generator according to the following pseudocode.
Pseudocode 75
for (ts = 0; ts < ts_offset_hfgen; ts++) { for (sb = 0; sb < num_qmf_subbands; sb++) { if (sb < sbx) Q_low[sb][ts] = Q_prev[sb][ts+num_qmf_timeslots-ts_offset_hfgen]; } } for (ts = ts_offset_hfgen; ts < (num_qmf_timeslots+ts_offset_hfgen); ts++) { for (sb = 0; sb < num_qmf_subbands; sb++) { if (sb < sbx) Q_low[sb][ts] = Q[sb][ts-ts_offset_hfgen]; } }
In this pseudocode, Q represents QinASPX, and Q_prev is the QinASPX matrix from the previous A-SPX interval. The crossover subband sbx is defined in clause 5.7.6.3.1.2.
The delay in QMF time slots, ts_offset_hfgen, can be derived from table 191.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)190
Table 191: Delay in A-SPX tool
frame_length num_ts_in_ats ts_offset_hfgen Delay in samples 2 048 2 6 3 * 2 * 64 = 384 1 920 2 6 3 * 2 * 64 = 384 1 536 2 6 3 * 2 * 64 = 384 1 024 1 3 3 * 1 * 64 = 192 960 1 3 3 * 1 * 64 = 192 768 1 3 3 * 1 * 64 = 192 512 1 3 3 * 1 * 64 = 192 384 1 3 3 * 1 * 64 = 192
5.7.6.3.3 Time/frequency matrix
5.7.6.3.3.0 Introduction
The A-SPX time/frequency matrix is derived from the QMF matrix Q_low. Similarly, it has time on the horizontal axis and frequency on the vertical axis. The A-SPX matrix is generated using two methods:
• framing, which determines the time slot borders of the signal and noise envelopes; and
• tiling, which determines the subband group structure for each of the envelopes.
Before the framing is applied, the QMF time slots are mapped to A-SPX time slots. Each A-SPX time slot (ats) consists of either one or two QMF time slots (ts), depending on the frame_length of the QMF analysis, as indicated by the num_ts_in_ats factor in table 191.
Pseudocode 75a
/* Conversion from QMF timeslots into A-SPX timeslots */ num_aspx_timeslots = num_qmf_timeslots / num_ts_in_ats;
5.7.6.3.3.1 Framing
The temporal extent of the time/frequency grouping for A-SPX in each frame is defined as the A-SPX interval. Four possible A-SPX interval classes exist, represented by the bitstream element aspx_int_class: FIXFIX, FIXVAR, VARFIX, and VARVAR.
The class name has two parts: the first denotes the nature of the start (left) border of the A-SPX interval, and the second denotes the nature of the stop (right) border of the A-SPX interval.
• FIX (fixed) means that either:
- the start border of the A-SPX interval coincides with QMF time slot 0 of Q_low; or the
- the stop border of the A-SPX interval coincides with the QMF time slot num_qmf_timeslots - 1 of Q_low.
• VAR (variable) means that the start or the stop border of the A-SPX interval does not coincide with either time slots mentioned above, but rather is variable.
Variable framing is used to allow for extending an A-SPX interval when a transient is situated at or very close to time slots representing the FIX borders. It allows a transient to be enclosed in its own envelope. For that, borders are shifted by an offset of 1 < ts_var_offset≤ts_offset_hfgen.
If an interval class name ends with FIX, it may only be followed by an interval class whose name starts with FIX, e.g. a FIXFIX interval may be followed either by a FIXFIX or a FIXVAR interval. Similarly, an interval class name ending with VAR may only be followed by an interval class whose name starts with VAR, e.g. a FIXVAR interval may only be followed either by a VARFIX or a VARVAR interval.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)191
The maximum size of an A-SPX interval is that of a FIXVAR interval with size num_qmf_timeslots + ts_offset_hfgen QMF time slots. The minimum size of an A-SPX interval is that of a VARFIX interval with size num_qmf_timeslots - ts_offset_hfgen QMF time slots.
Within this interval, the A-SPX time slots are grouped in A-SPX time slot groups (atsg), representing the signal and noise envelopes. The borders of these groups, represented by the arrays atsg_sig and atsg_noise, respectively, are derived according to the following pseudocode.
NOTE: For aspx_data_2ch elements with aspx_balance == 0, the framing needs to be calculated separately per channel. However, the channel index, ch, for the bitstream elements contained in aspx_framing is dropped in the following pseudocode in order to improve readability.
Pseudocode 76
num_atsg_sig = aspx_num_env; num_atsg_noise = aspx_num_noise; if (aspx_int_class == FIXFIX) { atsg_sig = tab_border[num_aspx_timeslots][num_atsg_sig]; atsg_noise = tab_border[num_aspx_timeslots][num_atsg_noise]; atsg_freqres[0] = freq_res(atsg_sig, 0, 0, num_aspx_timeslots, aspx_freq_res_mode); for (atsg = 1; atsg < num_atsg_sig; atsg++) atsg_freqres[atsg] = atsg_freqres[0]; } else { switch (aspx_int_class) { case FIXVAR: atsg_sig[0] = 0; atsg_sig[num_atsg_sig] = aspx_var_bord_right + num_aspx_timeslots; for (tsg = 0; tsg < aspx_num_rel_right; tsg++) atsg_sig[num_atsg_sig-tsg-1] = atsg_sig[num_atsg_sig-tsg] - aspx_rel_bord_right[tsg]; break; case VARFIX: if (b_iframe) atsg_sig[0] = aspx_var_bord_left; else atsg_sig[0] = previous_stop_pos - num_aspx_timeslots; atsg_sig[num_atsg_sig] = num_aspx_timeslots; for (tsg = 0; tsg < aspx_num_rel_left; tsg++) atsg_sig[tsg+1] = atsg_sig[tsg] + aspx_rel_bord_left[tsg]; break; case VARVAR: if (b_iframe) atsg_sig[0] = aspx_var_bord_left; else atsg_sig[0] = previous_stop_pos - num_aspx_timeslots; atsg_sig[num_atsg_sig] = aspx_var_bord_right + num_aspx_timeslots; for (tsg = 0; tsg < aspx_num_rel_left; tsg++) atsg_sig[tsg+1] = atsg_sig[tsg] + aspx_rel_bord_left[tsg]; for (tsg = 0; tsg < aspx_num_rel_right; tsg++) atsg_sig[num_atsg_sig-tsg-1] = atsg_sig[num_atsg_sig-tsg] - aspx_rel_bord_right[tsg]; break; } atsg_noise[0] = atsg_sig[0]; atsg_noise[num_atsg_noise] = atsg_sig[num_atsg_sig]; if (num_atsg_noise > 1) atsg_noise[1] = atsg_sig[noise_mid_border[aspx_tsg_ptr][aspx_int_class]]; for (atsg = 0; atsg < num_atsg_sig; atsg++) atsg_freqres[atsg] = freq_res(atsg_sig, atsg, aspx_tsg_ptr, num_aspx_timeslots, aspx_freq_res_mode); }
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)192
previous_stop_pos = atsg_sig[num_atsg_sig];
The tab_border array and noise_mid_border arrays and the function freq_res() are defined as follows.
Table 192: Index of the middle noise border (noise_mid_border)
aspx_tsg_ptr aspx_int_class VARFIX FIXVAR, VARVAR
-1 1 num_atsg_sig-1 ≥0 num_atsg_sig-1 max(1, min(num_atsg_sig-1, aspx_tsg_ptr))
Table 193: A-SPX time slot group borders for aspx_int_class FIXFIX (tab_border)
num_aspx_timeslots aspx_num_[env|noise] 1 2 4
6 {0, 6} {0, 3, 6} {0, 2, 3, 4, 6} 8 {0, 8} {0, 4, 8} {0, 2, 4, 6, 8} 12 {0, 12} {0, 6, 12} {0, 3, 6, 9, 12} 15 {0, 15} {0, 8, 15} {0, 4, 8, 12, 15} 16 {0, 16} {0, 8, 16} {0, 4, 8, 12, 16}
Pseudocode 77
freq_res(atsg_sig, atsg, aspx_tsg_ptr, num_aspx_timeslots, aspx_freq_res_mode) { switch (aspx_freq_res_mode) { case 0: freq_res = aspx_freq_res[atsg]; break; case 1: freq_res = 0; // FREQ_RES_LOW break; case 2: if ((atsg < aspx_tsg_ptr && num_aspx_timeslots > 8) || (atsg_sig[atsg+1]-atsg_sig[atsg]) > (num_aspx_timeslots/6.0+3.25)) freq_res = 1; // FREQ_RES_HIGH else freq_res = 0; // FREQ_RES_LOW break; case 3: freq_res = 1; // FREQ_RES_HIGH } return freq_res; }
The previous_stop_pos helper variable is used to store the stop border for next frame processing. It shall be initialized with num_aspx_timeslots.
5.7.6.3.3.2 Tiling
On the frequency axis, two subband group tables are available for signal envelopes, sbg_sig_highres and sbg_sig_lowres, and one for the noise envelopes, sbg_noise. They are derived in clause 5.7.6.3.1.2 and clause 5.7.6.3.1.3, respectively.
The resolution of the subband grouping for signal envelopes depends on atsg_freqres[atsg]. The subband grouping for an envelope represented by the A-SPX time slot group atsg is sbg_sig_highres for atsg_freqres[atsg] = 1, and sbg_sig_lowres if atsg_freqres[atsg] = 0.
In the following clauses, the subsequent mapping is used.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)193
Pseudocode 78
/* Mapping of high and low resolutions signal envelopes to time slot groups */ for (atsg = 0; atsg < num_atsg_sig; atsg++) { if (atsg_freqres[atsg]) { num_sbg_sig[atsg] = num_sbg_sig_highres; sbg_sig[atsg] = sbg_sig_highres; } else { num_sbg_sig[atsg] = num_sbg_sig_lowres; sbg_sig[atsg] = sbg_sig_lowres; } }
The subband grouping for any noise envelope is represented by sbg_noise.
Together, the time slot groups and the subband groups determine the time/frequency tiling for a given A-SPX interval.
5.7.6.3.4 Decoding A-SPX signal and noise envelopes
For each A-SPX signal envelope and noise envelope, the signal and noise scale factors are delta coded either along time or frequency directions. Upon crossing an A-SPX interval boundary (along the time axis), the first envelope in the current A-SPX interval may be delta coded using Huffman decoding with respect to the last envelope of the previous A-SPX interval. This is true for both signal and noise envelopes.
The following pseudocode describes how to choose the correct Huffman table for decoding the bitstream elements.
Pseudocode 79
get_aspx_hcb(data_type, quant_mode, stereo_mode, hcb_type) { // data_type = {SIGNAL, NOISE} // quant_mode = {0, 1} maps to qmode = {15, 30} indicating 1.5 dB or 3 dB // stereo_mode = {LEVEL, BALANCE} where 0 maps to LEVEL and 1 to BALANCE // hcb_type = {F0, DF, DT} if (data_type == SIGNAL) { aspx_hcb = ASPX_HCB_ENV_<stereo_mode>_<qmode>_<hcb_type>; // the line above expands using the inputs stereo_mode, qmode and hcb_type // example for the expansion (stereo_mode=LEVEL, qmode=15 and hcb_type=DF) // aspx_hcb = ASPX_HCB_ENV_LEVEL_15_DF; } else { // NOISE aspx_hcb = ASPX_HCB_NOISE_<stereo_mode>_<hcb_type>; } // The 18 A-SPX Huffman codebooks are given in table A.16 to table A.33 // and are named according to the scheme outlined above. return aspx_hcb; }
The quantized signal scale factors for each subband group, qscf_sig_sbg, are derived from the delta-coded signal scale factors aspx_data_sig as shown in the following pseudocode. Again, the channel index has been dropped from bitstream elements to improve readability.
Pseudocode 80
/* Index mapping sbg_sig_highres <-> sbg_sig_lowres */ sbg_idx_high2low[0] = 0; sbg_idx_low2high[0] = 0; sbg_low = 0; for (sbg = 0; sbg < num_sbg_sig_highres; sbg++) { if (sbg_sig_lowres[sbg_low+1] == sbg_sig_highres[sbg]) { sbg_low++; sbg_idx_low2high[sbg_low] = sbg; } sbg_idx_high2low[sbg] = sbg_low; } if ((ch == 1) && (aspx_balance == 1)) { delta = 2;
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)194
} else { delta = 1; } /* Loop over Envelopes */ for (atsg = 0; atsg < num_atsg_sig; atsg++) { /* Loop over scale factor subband groups */ for (sbg = 0; sbg < num_sbg_sig[tsg]); sbg++) { if (atsg == 0) { freq_res_prev[atsg] = atsg_freqres[num_atsg_sig_prev - 1]; qscf_prev[sbg][atsg] = qscf_sig_sbg_prev[sbg][num_atsg_sig_prev - 1]; } else { freq_res_prev[atsg] = atsg_freqres[atsg-1]; qscf_prev[sbg][atsg] = qscf_sig_sbg[sbg][atsg-1]; } if (aspx_sig_delta_dir[atsg] == 0) { /* FREQ */ qscf_sig_sbg[sbg][atsg] = 0; for (i = 0; i <= sbg; i++) { qscf_sig_sbg[sbg][atsg] += delta * aspx_data_sig[atsg][i]; } } else { /* TIME */ if (atsg_freqres[atsg] == freq_res_prev[atsg]) { qscf_sig_sbg[sbg][atsg] = qscf_prev[sbg][atsg]; qscf_sig_sbg[sbg][atsg] += delta * aspx_data_sig[atsg][sbg]); } else if ((atsg_freqres[atsg] == 0) && (freq_res_prev[atsg] == 1)) { qscf_sig_sbg[sbg][atsg] = qscf_prev[sbg_idx_low2high[sbg]][atsg]; qscf_sig_sbg[sbg][atsg] += delta * aspx_data_sig[atsg][sbg]; } else if ((atsg_freqres[atsg] == 1) && (freq_res_prev[atsg] == 0)) { qscf_sig_sbg[sbg][atsg] = qscf_prev[sbg_idx_high2low[sbg]][atsg]; qscf_sig_sbg[sbg][atsg] += delta * aspx_data_sig[atsg][sbg]; } } } }
The signal scale factors from the last envelope of the previous A-SPX interval, qscf_sig_sbg_prev, are needed when delta coding in the time direction over A-SPX interval boundaries. The number of signal envelopes of the previous A-SPX interval is denoted by num_atsg_sig_prev and is also needed in that case, as well as the frequency resolution vector of the previous A-SPX interval, denoted by atsg_freqres_prev.
The quantized noise envelope scale factors, qscf_noise_sbg, are derived from the delta-coded noise envelope data aspx_data_noise as shown in the following pseudocode.
Pseudocode 81
if ((ch == 1) && (aspx_balance == 1)) { delta = 2; } else { delta = 1; } /* Loop over envelopes */ for (atsg = 0; atsg < num_atsg_noise; atsg++) { /* Loop over noise subband groups */ for (sbg = 0; sbg < num_sbg_noise; sbg++) { qscf_noise_sbg[sbg][atsg] = 0; if (aspx_noise_delta_dir[atsg] == 0) { /* FREQ */ for (i = 0; i <= sbg; i++) { qscf_noise_sbg[sbg][atsg] += delta * aspx_data_noise[atsg][sbg]; } } else { /* TIME */ if (atsg == 0) { qscf_noise_sbg[sbg][atsg] = qscf_prev[sbg][num_atsg_noise_prev-1]; qscf_noise_sbg[sbg][atsg] += delta * aspx_data_noise[atsg][sbg]; } else {
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)195
qscf_noise_sbg[sbg][atsg] = qscf_noise_sbg[sbg][atsg-1]; qscf_noise_sbg[sbg][atsg] += delta * aspx_data_noise[atsg][sbg]; } } } }
In this pseudocode, qscf_prev is the noise envelope scale factors from last envelope of the previous A-SPX interval and num_atsg_noise_prev is the number of noise envelopes from the previous A-SPX interval.
5.7.6.3.5 Dequantization and stereo decoding
Two quantization steps are possible for the quantization of the signal scale factors:
• aspx_qmode_env[ch] = 0 corresponds to a quantization step of 1,5 dB; and
• aspx_qmode_env[ch] = 1 corresponds to a quantization step of 3,0 dB.
For a 1-channel element and for a 2-channel element with aspx_balance = 0, i.e. when stereo decoding is off, the quantized signal envelope scale factors, qscf_sig_sbg, are dequantized according to the following pseudocode.
Pseudocode 82
if (aspx_qmode_env[ch] == 0) a = 2; else a = 1; for (atsg = 0; atsg < num_atsg_sig; atsg++) { for (sbg = 0; sbg < num_sbg_sig[atsg]; sbg++) { scf_sig_sbg[sbg][atsg] = num_qmf_subbands * pow(2, qscf_sig_sbg[sbg][atsg]/a); } if (aspx_sig_delta_dir[atsg] == 0 && qscf_sig_sbg[0][atsg] == 0 && scf_sig_sbg[1][atsg] < 0) { scf_sig_sbg[0][atsg] = scf_sig_sbg[1][atsg]; } }
The noise envelope scale factors are dequantized as follows.
Pseudocode 83
NOISE_FLOOR_OFFSET = 6; for (atsg = 0; atsg < num_atsg_noise; atsg++) { for (sbg = 0; sbg < num_sbg_noise; sbg++) { scf_noise_sbg[sbg][atsg] = pow(2, NOISE_FLOOR_OFFSET - qscf_noise_sbg[sbg][atsg]); } }
The scf_sig_sbg and scf_noise_sbg factors are the dequantized signal and noise scale factors.
Decoding of two jointly coded A-SPX channels
If aspx_balance = 1, a channel pair is stereo coded as a sum and balance pair and they are jointly decoded. In that case, the time envelopes, atsg_sig and atsg_noise, are identical for the channels.
In the pseudocode below, scf_sig_sbg_a and scf_noise_sbg_a are the decoded signal and noise scale factors for the sum channel A and scf_sig_sbg_b and scf_noise_sbg_b are the decoded signal and noise scale factors for the balance channel B.
As output, the dequantized scale factors for signal and noise envelopes for channels A and B are retrieved.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)196
Pseudocode 84
NOISE_FLOOR_OFFSET = 6; PAN_OFFSET = 12; if (aspx_qmode_env[ch] == 0) a = 2; else a = 1; for (atsg = 0; atsg < num_atsg_sig; atsg++) { for (sbg = 0; sbg < num_sbg_sig[atsg]; sbg++) { nom = pow(2, qscf_sig_sbg_a[sbg][atsg]/a + 1) * num_qmf_subbands; denom_a = 1 + pow(2, PAN_OFFSET - qscf_sig_sbg_b[sbg][atsg]/a); denom_b = 1 + pow(2, qscf_sig_sbg_b[sbg][atsg]/a - PAN_OFFSET); scf_sig_sbg_a[sbg][atsg] = nom / denom_a; scf_sig_sbg_b[sbg][atsg] = nom / denom_b; } } for (atsg = 0; atsg < num_atsg_noise; atsg++) { for (sbg = 0; sbg < num_sbg_noise; sbg++) { nom = pow(2, NOISE_FLOOR_OFFSET - qscf_noise_sbg_a[sbg][atsg] + 1); denom_a = 1 + pow(2, PAN_OFFSET - qscf_noise_sbg_b[sbg][atsg]); denom_b = 1 + pow(2, qscf_noise_sbg_b[sbg][atsg] - PAN_OFFSET); scf_noise_sbg_a[sbg][atsg] = nom / denom_a; scf_noise_sbg_b[sbg][atsg] = nom / denom_b; } }
After the stereo decoding, both channels are processed similarly, but separately of each other. Hence, the channel indices are dropped, and the dequantized noise and envelope scale factors for each channel are represented by scf_noise_sbg and scf_sig_sbg.
5.7.6.4 HF signal construction
5.7.6.4.1 HF generator tool
5.7.6.4.1.1 Introduction
The HF generator patches subband signals from consecutive subbands of the matrix Q_low to consecutive subbands of the matrix Q_high based on parameters transmitted in the bitstream.
The tonal to noise ratio of the subband signals Q_high are adjusted to the levels signalled in the bitstream by the encoder.
Before the subband tonal to noise ratio adjustment, a pre-flattening step is performed by which a gain value is derived from a coarse approximation of the slope of the source range, Q_low, used for HF generation. The inverse of this gain value is applied during the patching process.
The noise and tone generator tools are used to add adaptive noise and sinusoids, at levels specified by the encoder, to the patched signals.
5.7.6.4.1.2 Pre-flattening control data calculation
The calculation of the pre-flattened control data involves the fitting of a third-order polynomial to the spectral envelope of the low band, Q_low. The fitted polynomial is a smoothed representation of the overall spectral slope of Q_low. The overall slope obtained is translated into a gain vector.
The pseudocode for the calculation of the pre-flattening control data is as follows.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)197
Pseudocode 85
mean_energy = 0; polynomial_order = 3; num_qmf_subbands = sbx; for (i = 0; i < num_qmf_subbands; i++) { x[i] = i; slope[i] = 0; } /* Calculate the spectral signal envelope in dB over the current interval. */ for (sb = 0; sb < num_qmf_subbands; sb++) { pow_env[sb] = 0; for (ts = atsg_sig[0]*num_ts_in_ats; ts < atsg_sig[num_atsg_sig]*num_ts_in_ats; ts++) { pow_env[sb] += pow(Q_low_real[sb][ts], 2) pow_env[sb] += pow(Q_low_imag[sb][ts], 2) } pow_env[sb] /= (atsg_sig[num_atsg_sig] - atsg_sig[0])*num_ts_in_ats; pow_env[sb] = 10*log10(pow_env[sb] + 1); mean_energy += pow_env[sb]; } mean_energy /= num_qmf_subbands; poly_array = polynomial_fit(polynomial_order, num_qmf_subbands, x, pow_env); /* Transform polynomial into slope */ for (k = polynomial_order; k >= 0; k--) { for (sb = 0; sb < num_qmf_subbands; sb++) { slope[sb] += pow(x[sb], k) * poly_array[polynomial_order - k]; } } /* Derive a gain vector from the slope */ for (sb = 0; sb < num_qmf_subbands; sb++) { gain_vec[sb] = pow(10, (mean_energy - slope[sb])/20); }
The polynomial_fit() function uses a least squares approach to fit to the spectral envelope of X_low. It is a standard function (see [i.12]) that returns an array of coefficients that define a polynomial that fits the energies in the subbands in a least-squares sense. The result, a vector poly_array of length 4, contains the polynomial coefficients in a descending order of powers.
5.7.6.4.1.3 Subband tonal to noise ratio adjustment data calculation
The purpose of the per subband tonal to noise ratio adjustment is to adjust the tonal to noise ratio within the subbands of the patched signal and this is a two-step process.
The first step is to perform linear prediction within each of the QMF subband signals of Q Low. The second step is the actual tonal to noise ratio adjustment, which is again performed independently for each of the subband signals patched to Q High by the HF generator, as described in clause 5.7.6.4.1.4.
The subband signals are complex valued, which results in:
• a complex covariance matrix, cov, for the linear prediction; and
• complex filter coefficients for the filter used to control the tonal to noise ratio.
The prediction filter coefficients are obtained using the covariance method. The covariance matrix elements are calculated according to the following pseudocode.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)198
Pseudocode 86
ts_offset_hfadj = 4; /* Create an additional delay of ts_offset_hfadj QMF time slots */ for (sb = 0; sb < sba; sb++) { ts_offset_prev = num_qmf_timeslots - ts_offset_hfadj; for (ts = 0; ts < ts_offset_hfadj; ts++) { Q_low_ext[sb][ts] = Q_low_prev[sb][ts+ts_offset_prev]; } for (ts = 0; ts < (num_qmf_timeslots+ts_offset_hfgen); ts++) { Q_low_ext[sb][ts+ts_offset_hfadj] = Q_low[sb][ts]; } } num_ts_ext = num_qmf_timeslots + ts_offset_hfgen+ts_offset_hfadj; /* Loop over QMF subbands */ for (sb = 0; sb < sba; sb++) { for (i = 0; i < 3; i++) { for (j = 1; j < 3; j++) { cov[sb][i][j] = 0; /* Loop over QMF time slots */ for (ts = ts_offset_hfadj; ts < num_ts_ext; ts+=2) { cov[sb][i][j] += Q_low_ext[sb][ts-2*i] * cplx_conj(Q_low_ext[sb][ts-2*j]); } } } }
The coefficient vectors alpha0 and alpha1 used to filter the subband signal are calculated as follows.
Pseudocode 87
EPSILON_INV = pow(2,-20); for (sb = 0; sb < sba; sb++) { denom = cov[sb][2][2] * cov[sb][1][1]; denom -= abs(cov[1][2]) * abs(cov[1][2]) * 1/(1+EPSILON_INV); if (denom == 0) { alpha1[sb] = 0; } else { alpha1[sb] = cov[sb][0][1] * cov[sb][1][2] - cov[sb][0][2] * cov[sb][1][1]; alpha1[sb] /= denom; } if (cov[sb][1][1] == 0) { alpha0[sb] = 0; } else { alpha0[sb] = - cov[sb][0][1] + alpha1[sb] * cplx_conj(cov[sb][1][2]); alpha0[sb] /= cov[sb][1][1]; } }
The pseudocode interpreter is assumed to handle complex arithmetic. The cplx_conj() function provides the complex conjugate of its parameter; abs() provides the magnitude of its parameter.
If either of the magnitudes of alpha0 and alpha1 is greater than or equal to 4, both coefficients are set to 0.
The amount of tonal to noise ratio adjustment is controlled by the values of the chirp factors, chirp_arr, which are calculated as shown in the following pseudocode. Each chirp factor is used within a specific frequency range defined by the noise envelope subband group table, sbg_noise.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)199
Pseudocode 88
for (sbg = 0; sbg < num_sbg_noise; sbg++) { new_chirp = tabNewChirp[aspx_tna_mode[sbg]][aspx_tna_mode_prev[sbg]]; if (new_chirp < prev_chirp_array[sbg]) { new_chirp = 0.75000 * new_chirp + 0.25000 * prev_chirp_array[sbg]; } else { new_chirp = 0.90625 * new_chirp + 0.09375 * prev_chirp_array[sbg]; } if (new_chirp < 0.015625) { chirp_arr[sbg] = 0; } else { chirp_arr[sbg] = new_chirp; } }
The prev_chirp_array[i] array contains the chirp factor values calculated in the previous A-SPX interval, which are assumed to be zero for the first A-SPX interval. The new_chirp values are derived by a lookup function that uses aspx_tna_mode_prev[i] and aspx_tna_mode[i] as inputs and that can be implemented using table 194. The aspx_tna_mode_prev[i] values are the aspx_tna_mode values from the previous A-SPX interval, and are assumed to be zero for the first interval.
Table 194: Calculation of new_chirp values
aspx_tna_mode_prev[i] aspx_tna_mode[i] None Light Moderate Heavy
None 0,0 0,6 0,9 0,98 Light 0,6 0,75 0,9 0,98 Moderate 0,0 0,75 0,9 0,98 Heavy 0,0 0,75 0,9 0,98
5.7.6.4.1.4 HF signal creation
The high-frequency signal, Q_high, is created using the following pseudocode, which implements both the tonal to noise ratio adjustment of the lower spectrum as well as the pre-flattening.
Pseudocode 89
/* Loop over QMF time slots */ for (ts = atsg_sig[0]*num_ts_in_ats; ts < atsg_sig[num_atsg_sig]*num_ts_in_ats; ts++) { sum_sb_patches = 0; g = 0; /* Loop over number of patches */ for (i = 0; i < num_sbg_patches; i++) { /* Loop over number of subbands per patch */ for (sb = 0; sb < sbg_patch_num_sb[i]; sb++) { /* Map to High QMF Subband */ sb_high = sbx + sum_sb_patches + sb; /* Map to current noise envelope */ if (sbg_noise[g+1] == sb_high) g++; n = ts + ts_offset_hfadj; /* Current low QMF Subband */ p = sbg_patch_start_sb[i] + sb; Q_high[sb_high][ts] = Q_low_ext[p][n]; Q_high[sb_high][ts] += chirp_arr[g] * alpha0[p] * Q_low_ext[p][n-2]; Q_high[sb_high][ts] += pow(chirp_arr[g],2) * alpha1[p] * Q_low_ext[p][n-4]; if (aspx_preflat == 1) Q_high[sb_high][ts] *= 1/gain_vec[p]; } sum_sb_patches += sbg_patch_num_sb[i]; } }
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)200
5.7.6.4.2 HF envelope adjustment tool
5.7.6.4.2.0 Introduction
The envelope adjustment process takes Q_high as input from the HF generator and produces as output a QMF matrix Y as output. The adjustment is performed upon the entire A-SPX range spanning the time slots of the current A-SPX interval (given by atsg_sig) and num_sb_aspx QMF subbands, starting at subband sbx.
5.7.6.4.2.1 Estimation of transmitted and actual envelopes in the current interval
The spectral envelopes within the current A-SPX interval are estimated depending on the bitstream element aspx_interpolation. The estimation is performed by taking the average of the squared complex subband samples over the time and frequency regions of the time-frequency matrix Q_high. The estimation can be calculated using the following pseudocode.
Pseudocode 90
/* Loop over envelopes */ for (atsg = 0; atsg < num_atsg_sig; atsg++) { sbg = 0; /* Loop over QMF subbands in A-SPX range */ for (sb = 0; sb < num_sb_aspx; sb++) { est_sig = 0; /* Update current subband group */ if (sb == sbg_sig[sbg+1]) sbg++; tsa = atsg_sig[atsg]*num_ts_in_ats + ts_offset_hfadj; tsz = atsg_sig[atsg+1]*num_ts_in_ats + ts_offset_hfadj; for (ts = tsa; ts < tsz; ts++) { if (aspx_interpolation == 0) { for (j = sbg_sig[sbg]; j < sbg_sig[sbg+1]; j++) { est_sig += pow(Q_high[j][ts], 2); } } else { est_sig += pow(Q_high[sb+sbx][ts], 2); } } if (aspx_interpolation == 0) { est_sig /= sbg_sig[sbg+1] - sbg_sig[sbg]; est_sig /= atsg_sig[atsg+1] - atsg_sig[atsg]; } else { est_sig /= atsg_sig[atsg+1] - atsg_sig[atsg]; } est_sig_sb[sb][atsg] = est_sig; } }
The frequency resolution of est_sig_sb equals that of the QMF filterbank. The est_sig_sb matrix has num_atsg_sig columns (one for every A-SPX envelope) and num_sb_aspx rows (the number of QMF subbands covered by the A-SPX range).
The Huffman decoded and dequantized signal and noise scale factors, scf_sig_sbg and scf_noise_sbg, are mapped to the resolution of the A-SPX time/frequency matrix, and the pseudocode governing the mapping is as follows.
Pseudocode 91
atsg_noise = 0; /* Loop over Signal Envelopes */ for (atsg = 0; atsg < num_atsg_sig; atsg++) { /* Map Signal Envelopes from subband groups to QMF subbands */ for (sbg = 0; sbg < num_sbg_sig; sbg++) { for (sb = sbg_sig[sbg]-sbx; sb < sbg_sig[sbg+1]-sbx; sb++) scf_sig_sb[sb][atsg] = scf_sig_sbg[sbg][atsg]; } if (atsg_sig[atsg] == atsg_noise[atsg_noise + 1])
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)201
atsg_noise++; /* Map Noise Floors from subband groups to QMF subbands, and to signal envelopes */ for (sbg = 0; sbg < num_sbg_noise; sbg++) { for (sb = sbg_noise[sbg]-sbx; sb < sbg_noise[sbg+1]-sbx; sb++) scf_noise_sb[sb][atsg] = scf_noise_sbg[sbg][atsg_noise]; } }
The following pseudocode shows how additional sinusoidal tones are added:
• sine_idx_sb is a binary matrix indicating the QMF subbands in which the sinusoids are to be added;
• sine_area_sb is a binary matrix indicating all QMF subbands of a subband group, into which a sinusoid is added.
The insertion of a sinusoid is signalled by the bitstream element aspx_add_harmonic. If a sinusoid is to be inserted in a QMF subband that, in the previous A-SPX interval, did not contain a sinusoid, the starting envelope of the inserted sinusoid is given by aspx_tsg_ptr in the present A-SPX interval. Next, the sinusoid is placed in the middle of the high-frequency resolution subband group. This is described by the following pseudocode.
Pseudocode 92
if (aspx_tsg_ptr_prev == num_atsg_sig_prev) p_sine_at_end = 0; else p_sine_at_end = -1; /* Loop over envelopes */ for (atsg = 0; atsg < num_atsg_sig; atsg++) { /* Loop over high resolution signal envelope subband groups */ for (sbg = 0; sbg < num_sbg_sig_highres; sbg++) { sba = sbg_sig_highres[sbg] - sbx; sbz = sbg_sig_highres[sbg+1] - sbx; sb_mid = (int) 0.5*(sbz+sba); /* Map sinusoid markers to QMF subbands */ for (sb = sbg_sig_highres[sbg]-sbx; sb < sbg_sig_highres[sbg+1]-sbx; sb++) { if ((sb == sb_mid) && ((atsg >= aspx_tsg_ptr) || (p_sine_at_end == 0) || sine_idx_sb_prev[sb][num_atsg_sig_prev-1])) { sine_idx_sb[sb][atsg] = aspx_add_harmonic[sbg]; } else { sine_idx_sb[sb][atsg] = 0; } } } }
The variables aspx_tsg_ptr_prev, sine_idx_sb_prev, and num_atsg_sig_prev correspond to aspx_tsg_ptr, sine_idx_sb, and num_atsg_sig of the previous A-SPX interval for the same subband range. If the subband range is larger for the current interval, the entries for the QMF subbands not covered by the previous sine_idx_sb are assumed to be zero.
The frequency resolution of signal scale factors varies and can either be coarse or fine. The frequency resolution of additionally generated sinusoids is always fine. The varying frequency resolution is handled as shown in the following pseudocode.
Pseudocode 93
/* Loop over Envelopes */ for (atsg = 0; atsg < num_atsg_sig; atsg++) { /* Loop over subband groups */ for (sbg = 0; sbg < num_sbg_sig[atsg]; sbg++) { b_sine_present = 0; /* Additional sinusoid present in SF band? */ for (sb = sbg_sig[atsg][sbg]-sbx; sb < sbg_sig[atsg][sbg+1]-sbx; sb++) { if (sine_idx_sb[sb][atsg] == 1) b_sine_present = 1;
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)202
} /* Mark all subbands in current subband group accordingly */ for (sb = sbg_sig[atsg][sbg]-sbx; sb < sbg_sig[atsg][sbg+1]-sbx; sb++) { sine_area_sb[sb][atsg] = b_sine_present; } } }
The matrix sine_area_sb is hence one for all QMF subbands in the subband groups where an additional sinusoid shall be added; it shall be 0 otherwise.
The resulting sinusoid and noise levels for each subband, sine_lev_sb and noise_lev_sb,, are calculated as follows.
Pseudocode 94
/* Loop over envelopes */ for (atsg = 0; atsg < num_atsg_sig; atsg++) { /* Loop over QMF subbands in A-SPX range */ for (sb = 0; sb < num_sb_aspx; sb++) { sig_noise_fact = scf_sig_sb[sb][atsg] / (1+scf_noise_sb[sb][atsg]); sine_lev_sb[sb][atsg] = sqrt(sig_noise_fact * sine_idx_sb[sb][atsg]); noise_lev_sb[sb][atsg] = sqrt(sig_noise_fact * scf_noise_sb[sb][atsg]); } }
5.7.6.4.2.2 Calculation of compensatory gains
In order for the QMF subband samples signals to retain the correct envelope, compensatory gains are calculated. The level of additional sinusoids as well as the level of the additional added noise are taken into account.
The initial values for the gain, sig_gain_sb, are derived from the dequantized scale factors for noise and signal envelopes, as well as from the estimated actual envelope.
Pseudocode 95
EPSILON = 1.0; if (aspx_tsg_ptr == num_atsg_sig) b_sine_at_end = 0; else b_sine_at_end = -1; /* Loop over envelopes */ for (atsg = 0; atsg < num_atsg_sig; atsg++) { /* Loop over QMF subbands in A-SPX range */ for (sb = 0; sb < num_sb_aspx; sb++) { if (sine_area_sb[sb][atsg] == 0) { denom = EPSILON + est_sig_sb[sb][atsg]; if (!(atsg == aspx_tsg_ptr || atsg == p_sine_at_end)) denom *= (1 + scf_noise_sb[sb][atsg]); sig_gain_sb[sb][atsg] = sqrt(scf_sig_sb[sb][atsg] / denom); } else { denom = EPSILON + est_sig_sb[sb][atsg]; denom *= 1 + scf_noise_sb[sb][atsg]; sig_gain_sb[sb][atsg] = sqrt(scf_sig_sb[sb][atsg]*scf_noise_sb[sb][atsg] / denom); } } }
The gain values are limited so as to avoid unwanted noise substitution. The following pseudocode shows the calculation of the maximum gain values used for the limiting.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)203
Pseudocode 96
LIM_GAIN = 1.41254; EPSILON0 = pow(10, -12); MAX_SIG_GAIN = pow(10, 5); /* Loop over envelopes */ for (atsg = 0; atsg < num_atsg_sig; atsg++) { /* Loop over limiter subband groups */ for (sbg = 0; sbg < num_sbg_lim; sbg++) { nom = 0; denom = EPSILON0; for (sb = sbg_lim[sbg]-sbx; sb < sbg_lim[sbg+1]-1-sbx; sb++) { nom += scf_sig_sb[sb][atsg]; denom += est_sig_sb[sb][atsg]; } max_sig_gain_sbg[sbg][atsg] = sqrt(nom/denom) * LIM_GAIN; } sbg = 0; /* Map to QMF subbands */ for (sb = 0; sb < num_sb_aspx; sb++) { if (sb == sbg_lim[sbg+1]-sbx) sbg++; max_sig_gain_sb[sb][atsg] = min(max_sig_gain_sbg[sbg][atsg], MAX_SIG_GAIN); } }
The pseudocode below shows how the additional noise added to the HF generated signal, noise_lev_sb, is limited in proportion to the energy lost when the gain values are limited.
Pseudocode 97
/* Loop over envelopes */ for (atsg = 0; atsg < num_atsg_sig; atsg++) { /* Loop over QMF subbands */ for (sb = 0; sb < num_sb_aspx; sb++) { tmp = noise_lev_sb[sb][atsg]; tmp *= max_sig_gain_sb[sb][atsg]/sig_gain_sb[sb][atsg]; noise_lev_sb_lim[sb][atsg] = min(noise_lev_sb[sb][atsg], tmp); } }
The compensatory gain values, sig_gain_sb, are limited as follows.
Pseudocode 98
/* Loop over envelopes */ for (atsg = 0; atsg < num_atsg_sig; atsg++) { /* Loop over QMF subbands */ for (sb = 0; sb < num_sb_aspx; sb++) { sig_gain_sb_lim[sb][atsg] = min(sig_gain_sb[sb][atsg], max_sig_gain_sb[sb][atsg]); } }
The total gain of a limiter subband group is adjusted in proportion to the energy lost during limiting. This boost factor is calculated as shown in the following pseudocode.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)204
Pseudocode 99
EPSILON0 = pow(10, -12); /* Loop over envelopes */ for (atsg = 0; atsg < num_atsg_sig; atsg++) { /* Loop over limiter subband groups */ for (sbg = 0; sbg < num_sbg_lim; sbg++) { nom = denom = EPSILON0; /* Loop over subbands */ for (sb = sbg_lim[sbg]-sbx; sb < sbg_lim[sbg+1]-1-sbx; sb++) { nom += scf_sig_sb[sb][atsg]; denom += est_sig_sb[sb][atsg]*pow(sig_gain_sb_lim[sb][atsg], 2); denom += pow(sine_lev_sb[sb][atsg], 2); if (!((sine_lev_sb[sb][atsg] != 0) || (atsg == aspx_tsg_ptr) || (atsg == p_sine_at_end))) denom += pow(noise_lev_sb_lim[sb][atsg], 2); } boost_fact_sbg[sbg][atsg] = sqrt(nom/denom); } }
The actual boost factor, boost_fact_sb, is limited in order not to get too high energy values as follows.
Pseudocode 100
MAX_BOOST_FACT = 1.584893192; /* Loop over envelopes */ for (atsg = 0; atsg < num_atsg_sig; atsg++) { sbg = 0; /* Loop over QMF subbands */ for (sb = 0; sb < num_sb_aspx; sb++) { if (sb == sbg_lim[sbg+1]-sbx) sbg++; boost_fact_sb[sb][atsg] = min(boost_fact_sbg[sbg][atsg], MAX_BOOST_FACT); } }
The boost factor is then applied to the compensation gain, the noise envelope scale factors and the sinusoid levels as shown in the following pseudocode.
Pseudocode 101
/* Loop over envelopes */ for (atsg = 0; atsg < num_atsg_sig; atsg++) { /* Loop over QMF subbands */ for (sb = 0; sb < num_sb_aspx; sb++) { boost_fact = boost_fact_sb[sb][atsg]; sig_gain_sb_adj[sb][atsg] = sig_gain_sb_lim[sb][atsg] * boost_fact; noise_lev_sb_adj[sb][atsg] = noise_lev_sb_lim[sb]atsg] * boost_fact; sine_lev_sb_adj[sb][atsg] = sine_lev_sb[sb][atsg] * boost_fact; } }
5.7.6.4.3 Noise generator tool
Overview
The noise generator tool is used to inject noise into the recreated high-frequency spectrum in order to decrease its tonality and to match the original signal more closely.
The tool outputs a noise envelope that is subsequently added to the generated HF signal.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)205
Data and control interfaces
Input:
• noise_lev_sb_adj, a num_sb_aspx x num_atsg_noise matrix of noise levels.
• NoiseTable, containing 512 complex numbers with random phase and average energy of 1, as defined in table D.2.
Output:
• qmf_noise, a num_sb_aspx x num_qmf_timeslots matrix of noise values.
Operation
The noise to be added to the recreated high-frequency subbands, qmf_noise, is generated according to the following pseudocode.
Pseudocode 102
/* Loop over time slots */ atsg = 0; for (ts = atsg_sig[0]*num_ts_in_ats; ts < atsg_sig[num_atsg_sig]*num_ts_in_ats; ts++) { if (ts == atsg_sig[atsg+1]*num_ts_in_ats) atsg++; /* Loop over QMF subbands in A-SPX */ for (sb = 0; sb < num_sb_aspx; sb++) { qmf_noise[sb][ts] = noise_lev_sb_adj[sb][atsg] * NoiseTable[noise_idx(sb,ts)]; } }
The index into the noise table is calculated according to the following pseudocode.
Pseudocode 103
noise_idx(sb, ts) { if (master_reset) { indexNoise = 0; } else { indexNoise = noise_idx_prev[sb][ts]; } indexNoise += num_sb_aspx * (ts - atsg_sig[0]); indexNoise += sb + 1; return indexNoise % 512; }
The noise_idx_prev variable is the last noise_idx from the previous A-SPX interval. The variable master_reset is defined in clause 5.7.6.3.1.1.
5.7.6.4.4 Tone generator tool
Overview
The tone generator tool is used to match the tonality of the original signal by adding the appropriate missing sinusoids to the recreated high band.
The tool outputs an envelope worth of values, the dimensions of which are specified in the bitstream by the encoder.
Data and control interfaces
Input:
• sine_lev_sb_adj, a num_sb_aspx x num_atsg_sig matrix of boosted sine levels.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)206
Output:
• qmf_sine, a num_sb_aspx x num_qmf_timeslots matrix of sine tones.
Operation
The sinusoids are added at the level sine_lev_sb_adj for the QMF subbands. This results in the final output QMF matrix, qmf_sine, calculated as follows.
Pseudocode 104
atsg = 0; /* Loop over QMF time slots */ for (ts = atsg_sig[0]*num_ts_in_ats; ts < atsg_sig[num_atsg_sig]*num_ts_in_ats; ts++) { if (ts == atsg_sig[atsg+1]*num_ts_in_ats) atsg++; /* Loop over QMF subbands in A-SPX */ for (sb = 0; sb < num_sb_aspx; sb++) { qmf_sine_RE[sb][ts] = sine_lev_sb_adj[sb][atsg]; qmf_sine_RE[sb][ts] *= SineTable_RE[sine_idx(sb, ts)]; qmf_sine_IM[sb][ts] = sine_lev_sb_adj[sb][atsg] * pow(-1, sb+sbx); qmf_sine_IM[sb][ts] *= SineTable_IM[sine_idx(sb, ts)]; } }
The index into the sine table is calculated according to the following pseudocode.
Pseudocode 105
sine_idx(sb, ts) { if (first_frame) { index = 1; } else { index = (sine_idx_prev[sb][ts] + 1) % 4; } index += ts - atsg_sig[0]; return index % 4; }
The sine_idx_prev variable is the last sine_idx from the previous A-SPX interval. The variable first_frame is 1 only at the codec initialization stage; 0 otherwise.
The array SineTable is defined in table 195.
Table 195: SineTable values for tone generator tool
index SineTable_RE(index) SineTable_IM(index) 0 1 0 1 0 1 2 -1 0 3 0 -1
5.7.6.4.5 HF signal assembling tool
The compensated gain values are applied to the input subband matrix, Q_high, for all signal envelopes of the current A-SPX interval as follows.
NOTE: For variable A-SPX interval borders, the previously assembled HF signals, Y_prev, are pushed into the current output buffer Y.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)207
Pseudocode 106
atsg = 0; /* Get delayed QMF subsamples from delay buffer */ for (ts = 0; ts < atsg_sig[0]*num_ts_in_ats; ts++) for (sb = 0; sb < num_sb_aspx; sb++) { Y[sb][ts] = Y_prev[sb][num_qmf_timeslots+ts]; } } /* Loop over QMF time slots */ for (ts = atsg_sig[0]*num_ts_in_ats; ts < atsg_sig[num_atsg_sig]*num_ts_in_ats; ts++) { if (ts == atsg_sig[atsg+1]*num_ts_in_ats) atsg++; /* Loop over QMF subbands */ for (sb = 0; sb < num_sb_aspx; sb++) { Y[sb][ts] = sig_gain_sb_adj[sb][atsg]; Y[sb][ts] *= Q_high[sb+sbx][ts+ts_offset_hfadj]; } }
The noise level, noise_lev_sb_adj, is passed as an input to the noise generator tool, and the generated noise qmf_noise is then added to the matrix Y as follows.
Pseudocode 107
/* Loop over time slots */ for (ts = atsg_sig[0]; ts < atsg_sig[num_atsg_sig]; ts++) { /* Loop over QMF subbands */ for (sb = 0; sb < num_sb_aspx; sb++) { Y[sb][ts] = Y[sb][ts] + qmf_noise[sb][ts]; } }
Similarly, the level of the sinusoids, sine_lev_sb_adj, is passed as an input to the tone generator tool and the resultant sinusoids qmf_sine are added to the matrix Y as follows.
Pseudocode 108
/* Loop over time slots */ for (ts = atsg_sig[0]; ts < atsg_sig[num_atsg_sig]; ts++) { /* Loop over QMF subbands */ for (sb = 0; sb < num_sb_aspx; sb++) { Y[sb][ts] = Y[sb][ts] + qmf_sine[sb][ts]; } }
5.7.6.5 Interleaved waveform coding
5.7.6.5.1 Introduction
A-SPX involves patching the spectral content of suitable low-frequency regions to higher frequencies and adjusting them so that they perceptually match the high-frequency content that was present in the original signal. The entire signal spectrum above the A-SPX crossover frequency is derived from the spectrum below the crossover frequency, and is augmented by synthetic signal components. noise and sinusoids, that are created using a parametric model.
The AC-4 encoder uses interleaved waveform coding either when critical signal frequency components are not present below the A-SPX crossover frequency, or when more faithful reproduction of transients is needed. Waveform coded components (WCC) and spectral extension components (SEC) can be interleaved either in frequency or in time. The signal characteristics determine whether frequency interleaving or time interleaving is more beneficial.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)208
Signals with a significant degree of tonal high-frequency content will profit from waveform-coding spectral lines that contain the tonal components and interleaving them with the spectrum created by spectral extension. If frequency interleaving is signalled in the bitstream, the WCC are added to the SEC.
On the other hand, the higher frequencies of a transient event can be coded more accurately by a short segment of waveform coding and interleaved in time with a spectrum generated by spectral extension. In the case of time interleaving, the WCC are continuous frequency regions that match the QMF subband structure. Hence the method for time interleaved coding is replacing the SEC with the corresponding WCC rather than addition. The resolution of time-interleaved waveform coding is the stride factor (num_ts_in_ats).
5.7.6.5.2 Signalling interleaved waveform coding
The use of frequency-interleaved waveform coding in an A-SPX interval is indicated by the bitstream element aspx_fic_present. The presence of a frequency-interleaved waveform component in a particular subband group is indicated by the bitstream element aspx_fic_used_in_sfb.
For channel pairs, the bitstream elements aspx_fic_left and aspx_fic_right indicate whether or not frequency-interleaved waveform coding is used in the left and right channels respectively. The bitstream element aspx_fic_used_in_sfb takes an additional channel index.
The use of time-interleaved waveform coding, starting at the A-SPX timeslot that coincides with QMF timeslot 0, is indicated by the bitstream element aspx_tic_present. The presence of a time-interleaved waveform component in a particular QMF time slot is indicated by the bitstream element aspx_tic_used_in_slot.
For channel pairs, the bitstream elements aspx_tic_left and aspx_tic_right indicate whether or not time-interleaved waveform coding is used in the left and right channels respectively. The bitstream element aspx_tic_used_in_slot takes an additional channel index. If the bitstream element aspx_tic_copy is set, information regarding the presence of time-interleaved waveform components in the slots of the left channel is copied to the right channel, resulting in both channels having time-interleaved components in the same slots.
The relative priorities of frequency-interleaving, time-interleaving and sinusoid synthesis (using the tone generator tool) are in the following descending order of priority:
1) Time-interleaved coding overrides both frequency-interleaved coding and sinusoid synthesis.
2) Sinusoid synthesis overrides frequency-interleaved coding; however, the sinusoid starts after a transient if a transient is present in the A-SPX interval. In this case, waveform coding followed by a synthetic sinusoid in the same band may be present.
3) Frequency-interleaved coding.
It should be noted that frequency-interleaved WCC are not required to cover the entire QMF subband group in which they are present.
5.7.6.5.3 Interleaving WCC and SEC
The WCC components are contained in the matrix QinASPX. The SEC are contained in the matrix Y.
For frequency-interleaved waveform coding, the WCC are added to the SEC as follows:
; Re{QoutASPX�m,i)< =Re{QinASPX�m,i-δASPX)< + ���Y�m, i�� Im{QoutASPX�m,i)� = ��{QinASPX�m,i-δASPX)�+ ��{�m,i)�
For time-interleaved waveform coding, the WCC replaces the SEC as follows:
� Re{QoutASPX�m,i)� =Re�QinASPX�m, i-δASPX� Im{QoutASPX�m,i)� = ��{QinASPX�m,i-δASPX)�
where δASPX=ts_offset_hfgen is the overall delay introduced by A-SPX processing.
The output matrix, QoutASPX, is input to the downstream QMF domain processing tool, as depicted in figure 6.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)209
5.7.7 Advanced coupling tool - A-CPL
5.7.7.1 Introduction
The A-CPL tool is a coding tool for improved coding of signals with more than one channel.
The tool operates in one of the following configurations:
Advanced coupling in the channel pair element For improved stereo coding. This configuration is described in clause 5.7.7.5.
Advanced coupling in the multichannel element
For improved multichannel coding. This configuration is described in clause 5.7.7.6.
Data and control interfaces
Input:
QinACPL,[a|b|…] MACPL complex QMF matrices of MACPL channels to be processed by the A-CPL tool.
Output:
QoutACPL,[a|b|…] MACPL complex spectrally decoupled QMF matrices corresponding to the same channels.
The QinACPL and QoutACPL matrices each consist of num_qmf_timeslots columns and num_qmf_subbands rows. MACPL is the number of channels contained in the channel element with the exception of the LFE.
Control:
• Decoded and dequantized advanced coupling parameters.
Figure 7 illustrates the interconnection of the advanced coupling tool with the other AC-4 coding tools.
Figure 7: Advanced Coupling block diagram
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)210
The input to the advanced coupling tool corresponds to the output of the A-SPX tool. The advanced coupling tool consists of one or more parallel modules that generically take two channels as input and provide one-channel or two-channel output. The two-channel input to the advanced coupling tool is subjected to a two-channel input conversion that provides a single modified signal by means of a non-linear decorrelation operation. The two-channel output from the advanced coupling tool module is formed from the two-channel input, the modified input signal, and the corresponding bitstream elements.
5.7.7.2 Parameter band to QMF subband mapping
Advanced Coupling parameters are transmitted per parameter band. Parameter bands are groupings of QMF subbands and have lower frequency resolution than the QMF subbands. The mapping of parameter bands to QMF subbands is shown below in table 196. The number of parameter bands, 7, 9, 12, or 15, is given by the variable acpl_num_param_bands.
Table 196: Mapping of parameter bands to QMF subbands
QMF band groups acpl_num_param_bands 15 12 9 7
0 0 0 0 0 1 1 1 1 1 2 2 2 2 2 3 3 3 3 2 4 4 4 3 3 5 5 4 4 3 6 6 5 4 3 7 7 5 5 3 8 8 6 5 4 9 - 10 9 6 6 4 11 - 13 10 7 6 4 14 - 17 11 8 7 5 18 - 22 12 9 7 5 23 - 34 13 10 8 6 35 - 63 14 11 8 6
5.7.7.3 Interpolation
Decoded and dequantized advanced coupling parameters or products of these (following advanced coupling parameters) carried in the bitstream are time-interpolated to calculate values that are applied to the input of the decorrelator and to the ducked output of the decorrelator.
Parameter sets are transmitted either once or twice per frame, which is determined by the variable acpl_num_param_sets.
Two forms of interpolation, smooth and steep, are utilized to interpolate values for each QMF subsample.
When smooth interpolation is used, the values for each QMF subsample between consecutive parameter sets are linearly interpolated.
When steep interpolation is used, the QMF subsamples at which parameter set values change are indicated by the bitstream element acpl_param_timeslot.
The interpolate() function, which is used in clause 5.7.7.5 and clause 5.7.7.6, is given by the following pseudocode.
Pseudocode 109
interpolate(acpl_param, num_pset, sb, ts) { num_ts = num_qmf_timeslots; if (acpl_interpolation_type == 0) { // smooth interpolation if (num_pset == 1) { // 1 parameter set delta = acpl_param[0][sb_to_pb(sb)] - acpl_param_prev[sb]; interp = acpl_param_prev[sb] + (ts+1)*delta/num_ts; }
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)211
else { // 2 parameter sets ts_2 = floor(num_ts/2); if (ts < ts_2) { delta = acpl_param[0][sb_to_pb(sb)] - acpl_param_prev[sb]; interp = acpl_param_prev[sb] + (ts+1)*delta/ts_2; } else { delta = acpl_param[1][sb_to_pb(sb)] - acpl_param[0][sb_to_pb(sb)]; interp = acpl_param[0][sb_to_pb(sb)] + (ts-ts_2+1)*delta/(num_ts-ts_2); } } } else { // steep interpolation if (num_pset == 1) { // 1 parameter set if (ts < acpl_param_timeslot[0]) { interp = acpl_param_prev[sb]; } else { interp = acpl_param[0][sb_to_pb(sb)]; } } else { // 2 parameter sets if (ts < acpl_param_timeslot[0]) { interp = acpl_param_prev[sb]; } else if (ts < acpl_param_timeslot[1]) { interp = acpl_param[0][sb_to_pb(sb)]; } else { interp = acpl_param[1][sb_to_pb(sb)]; } } } return interp; }
The sb_to_pb() function maps from QMF subbands to parameter bands according to table 196. The acpl_param_prev[sb] array, which holds the dequantized advanced coupling parameters from the previous AC-4 frame related to the provided acpl_param[pset][pb] array, is also passed on to the interpolate() function although acpl_param_prev is not shown as input parameter. The initialization of acpl_param_prev[sb] for all relevant dequantized advanced coupling parameter arrays for the next AC-4 frame is done at the end of the A-CPL tool processing for the current frame as shown in the following pseudocode.
Pseudocode 110
for (sb = 0; sb < num_qmf_subbands; sb++) { acpl_param_prev[sb] = acpl_param[num_pset-1][sb_to_pb(sb)]; }
When decoding the first AC-4 frame, all elements of all acpl_param_prev[sb] arrays shall be 0.
5.7.7.4 Decorrelator and transient ducker
5.7.7.4.1 Introduction
The decorrelation filters consist of a number of all-pass IIR filter sections preceded by a constant frequency-dependent delay. The frequency axis of the QMF time/frequency matrix is divided into three different regions as shown in table 197, which correspond to the QMF split frequencies given in clause 5.7.7.2. The length of the delay and the length of the filter coefficients vectors are identical within each region.
Table 197: Division of QMF subbands
Subband region Subbands Delay Filter length k0 0-6 7 7 k1 7-22 10 4 k2 23-63 12 2
The filtered signal is then ducked to lower the level of the reverberation tails.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)212
5.7.7.4.2 Decorrelator IIR filtering
The decorrelator filters for the different frequency regions given in table 197 are implemented by a delay and a subsequent IIR filter.
Pseudocode 111
inputSignalModification(x) { for (sb = 0; sb < num_qmf_subbands; sb++) { for (ts = 0; ts < num_qmf_timeslots; ts++) { b[0] = a[filterLength(sb)]; y[ts][sb] = b[0]*x[ts-delay(sb)][sb]/a[0]; for (i = 1; i <= filterLength(sb); i++) { b[i] = a[filterLength(sb)-i]; y[ts][sb] += (b[i]*x[ts-i-delay(sb)][sb] - a[i]*y[ts-i][sb])/a[0]; } } } return y; }
NOTE: Negative indices in the pseudocode indicate access to filter states from previously processed frames.
Three decorrelators, D0, D1, and D2, are available. Decorrelator Dx is used for the calculation of the intermediate signal named ux. The coefficients a[i] for the different regions are given in table 198, table 199, and table 200, and the coefficients b[i] can be derived from those using the pseudocode above.
Table 198: Coefficients a[i] for region k0
i D0 D1 D2 0 1,0000 1,0000 1,0000 1 0,5306 -0,4178 0,4007 2 -0,4533 0,1082 0,4747 3 -0,6248 -0,2368 0,2611 4 0,0424 -0,1014 -0,1211 5 0,4237 -0,1052 -0,4248 6 0,4311 -0,3528 -0,2989 7 0,1688 0,4665 -0,1932
Table 199: Coefficients a[i] for region k1
i D0 D1 D2 0 1,0000 1,0000 1,0000 1 0,5561 0,0425 -0,4361 2 -0,3039 0,3235 0,0345 3 -0,5024 -0,1556 0,5215 4 -0,1850 0,4958 -0,4178
Table 200: Coefficients a[i] for region k2
i D0 D1 D2 0 1,0000 1,0000 1,0000 1 0,5773 0,2327 -0,6057 2 0,3321 -0,3901 0,3804
5.7.7.4.3 Transient ducker
To be able to handle transients and other fast time-envelopes, the output of the decorrelator all-pass filter has to be attenuated and this is performed according to the following pseudocode.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)213
Pseudocode 112
ALPHA = 0.76592833836465; ALPHA_SMOOTH = 0.25; GAMMA = 1.5; EPSILON = 1.0e-9; acpl_max_num_param_bands = 15; for (pb = 0; pb < acpl_max_num_param_bands; pb++) { if (ALPHA * p_peak_decay_prev[pb] < p_energy[pb]) { p_peak_decay[pb] = p_energy[pb]; } else { p_peak_decay[pb] = ALPHA * p_peak_decay_prev[pb]; } smooth[pb] = (1.0f – ALPHA_SMOOTH) * p_smooth_prev[pb]; smooth[pb] += ALPHA_SMOOTH * p_energy[pb]; smooth_peak_diff[pb] = (1.0f – ALPHA_SMOOTH) * smooth_peak_diff_prev[pb]; smooth_peak_diff[pb] += ALPHA_SMOOTH * (p_peak_decay[pb] - p_energy[pb]); if (GAMMA * smooth_peak_diff[pb] > smooth[pb]) { duck_gain[pb] = smooth[pb] / (GAMMA * (smooth_peak_diff[pb] + EPSILON)); } else { duck_gain[pb] = 1.0f; } }
In this pseudocode, the arrays p_smooth_prev, p_peak_decay_prev and p_smooth_peak_diff_prev are the p_smooth, p_peak_decay and p_smooth_peak_diff arrays of the previous frame, respectively, and the array p_energy contains the energy per parameter band of the input channel.
NOTE: When there is no previous state, e.g. at startup, these arrays should be set to 0.
The value for p_energy can be calculated as follows.
Pseudocode 113
for (pb=0; pb < acpl_max_num_param_bands; pb++) { p_energy[pb] = 0; for (sb=first_sb_in_pb; sb <= last_sb_in_pb; sb++) { p_energy[pb] += pow(abs(x[sb]), 2); } }
The first_sb_in_pb variable is the first QMF band, and the last_sb_in_pb variable is the last QMF band in the corresponding parameter band, as specified in clause 5.7.7.2. x[sb] denotes the QMF subsample of the corresponding QMF band sb.
The transient attenuation value, duck_gain[pb], is applied to the output of the decorrelator according to the following pseudocode.
Pseudocode 114
applyTransientDucker(x) { for (sb = 0; sb < num_qmf_subbands; sb++) { pb = sb_to_pb(sb); y[sb] = x[sb] * duck_gain[pb]; } return y; }
The sb_to_pb() function maps from QMF subbands to parameter bands according to table 196.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)214
5.7.7.5 Advanced coupling in the channel pair element
For advanced coupling in the channel pair element, two input channels are present. Their elements are addressed as follows:
Left input channel x 0(ts,sb) ∈ Qin ACPL,L
Right input channel
x 1(ts,sb) ∈ Qin ACPL,R
The output channels are addressed as:
Left output channel z 0(ts,sb) ∈ Qout ACPL,L
Right output channel
z 1(ts,sb) ∈ Qout ACPL,R
They are derived according to the following pseudocode.
Pseudocode 115
x0in = 2*x0; u0 = inputSignalModification(x0in); // use decorrelator D0 y0 = applyTransientDucker(u0); num_pset = acpl_num_param_sets; if (stereo_codec_mode == ASPX_ACPL_1) { x1in = 2*x1; (z0, z1) = ACplModule(acpl_alpha_1_dq, acpl_beta_1_dq, num_pset, x0in, x1in, y0); } else if (stereo_codec_mode == ASPX_ACPL_2) { (z0, z1) = ACplModule(acpl_alpha_1_dq, acpl_beta_1_dq, num_pset, x0in, 0, y0); }
The inputSignalModification() and applyTransientDucker() functions are outlined in clause 5.7.7.4.2 and clause 5.7.7.4.3 respectively, and ACplModule() is defined as:
Pseudocode 116
ACplModule(acpl_alpha, acpl_beta, num_pset, x0, x1, y) { for (sb = 0; sb < num_qmf_subbands; sb++) {
for (ts = 0; ts < num_qmf_timeslots; ts++) {
interp_a = interpolate(acpl_alpha, num_pset, sb, ts); interp_b = interpolate(acpl_beta, num_pset, sb, ts);
if (sb < acpl_qmf_band) { z0[ts][sb]= 0.5*(x0[ts][sb] + x1[ts][sb]); z1[ts][sb]= 0.5*(x0[ts][sb] - x1[ts][sb]); } else { z0[ts][sb]= 0.5*(x0[ts][sb]*(1+interp_a) + y[ts][sb]*interp_b); z1[ts][sb]= 0.5*(x0[ts][sb]*(1-interp_a) - y[ts][sb]*interp_b); }
} } }
Here, interpolate() implements the interpolation outlined in clause 5.7.7.3. The arrays acpl_alpha_1_dq and acpl_beta_1_dq are derived from acpl_alpha1 and acpl_beta1 by differential decoding and dequantization as described in clause 5.7.7.7.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)215
5.7.7.6 Advanced coupling in the multichannel element
5.7.7.6.1 5_X_codec_mode ∈ {ASPX_ACPL_1, ASPX_ACPL_2}
For the configuration used when 5_X_codec_mode = ASPX_ACPL_1 or ASPX_ACPL_2, two parallel Advanced Coupling modules are used.
Five input channels are present. Their elements are addressed as:
Left input channel x0(ts,sb) ∈ QinACPL,L
Right input channel
x1(ts,sb) ∈ QinACPL,R Centre input channel
x2(ts,sb) ∈ QinACPL,C Left surround input channel
x3(ts,sb) ∈ QinACPL,Ls Right surround input channel
x4(ts,sb) ∈ QinACPL,Rs
The output channels are addressed as:
Left output channel z0(ts,sb) ∈ QoutACPL,L
Right output channel
z2(ts,sb) ∈ QoutACPL,R Centre output channel
z4(ts,sb) ∈ QoutACPL,C Left surround output channel
z1(ts,sb) ∈ QoutACPL,Ls Right surround output channel
z3(ts,sb) ∈ QoutACPL,Rs
They are derived according to the following pseudocode.
Pseudocode 117
x0in = 2*x0; x1in = 2*x1; u0 = inputSignalModification(x0in); // use decorrelator D0 u1 = inputSignalModification(x1in); // use decorrelator D1 y0 = applyTransientDucker(u0); y1 = applyTransientDucker(u1); if (5_X_codec_mode == ASPX_ACPL_1) { x3in = 2*x3; x4in = 2*x4; (z0, z1) = ACplModule(acpl_alpha_1_dq, acpl_beta_1_dq, num_pset_1, x0in, x3in, y0); (z2, z3) = ACplModule(acpl_alpha_2_dq, acpl_beta_2_dq, num_pset_2, x1in, x4in, y1); } else if (5_X_codec_mode == ASPX_ACPL_2) { (z0, z1) = ACplModule(acpl_alpha_1_dq, acpl_beta_1_dq, num_pset_1, x0in, 0, y0); (z2, z3) = ACplModule(acpl_alpha_2_dq, acpl_beta_2_dq, num_pset_2, x1in, 0, y1); } z1 *= sqrt(2);
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)216
z3 *= sqrt(2);
z4 = x2;
The functions inputSignalModification() and applyTransientDucker() are outlined in clause 5.7.7.4.2 and clause 5.7.7.4.3, respectively, and ACplModule() is defined in clause 5.7.7.5.
The variables num_pset_1 and num_pset_2 indicate the value acpl_num_param_sets of the first and the second acpl_data_1ch() element, respectively.
The arrays acpl_alpha_1_dq and acpl_beta_1_dq are derived from acpl_alpha1 and acpl_beta1 of the first acpl_data_1ch() element and the arrays acpl_alpha_2_dq and acpl_beta_2_dq are derived from acpl_alpha1 and acpl_beta1 of the second acpl_data_1ch() element by differential decoding and dequantization as described in clause 5.7.7.7.
5.7.7.6.2 5_X_codec_mode = ASPX_ACPL_3
For the configuration used when 5_X_codec_mode = ASPX_ACPL_3, three parallel advanced coupling modules are used.
Five input channels are present. Their elements are addressed as:
Left input channel x0(ts,sb) ∈ QinACPL,L
Right input channel
x1(ts,sb) ∈ QinACPL,R
Centre input channel
x2(ts,sb) ∈ QinACPL,C
Left surround input channel
x3(ts,sb) ∈ QinACPL,Ls
Right surround input channel
x4(ts,sb) ∈ QinACPL,Rs
The output channels are addressed as:
Left output channel z0(ts,sb) ∈ QoutACPL,L
Right output channel
z2(ts,sb) ∈ QoutACPL,R
Centre output channel
z4(ts,sb) ∈ QoutACPL,C
Left surround output channel
z1(ts,sb) ∈ QoutACPL,Ls
Right surround output channel
z3(ts,sb) ∈ QoutACPL,Rs
They are derived according to the following pseudocode.
Pseudocode 118
x0in = x0*(1+2*sqrt(0.5)); x1in = x1*(1+2*sqrt(0.5)); num_pset = acpl_num_param_sets; v1 = Transform(g1_dq, g2_dq, num_pset, x0in, x1in); v2 = Transform(g3_dq, g4_dq, num_pset, x0in, x1in); v3 = Transform(g1_dq + g3_dq + g5_dq, g2_dq + g4_dq + g6_dq, num_pset, x0in, x1in);
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)217
u0 = inputSignalModification(v1); // use decorrelator D0 u1 = inputSignalModification(v2); // use decorrelator D1 u2 = inputSignalModification(v3); // use decorrelator D2 y0 = applyTransientDucker(u0); y1 = applyTransientDucker(u1); y2 = applyTransientDucker(u2); (z0, z1) = ACplModule2(g1_dq, g2_dq, acpl_alpha_1_dq, acpl_beta_1_dq, num_pset, x0in, x1in, y0); (z2, z3) = ACplModule2(g3_dq, g4_dq, acpl_alpha_2_dq, acpl_beta_2_dq, num_pset, x0in, x1in, y1); (z4, z5) = ACplModule2(g5_dq, g6_dq, 1, 0, num_pset, x0in, x1in, 0); (z0, z1) = ACplModule3(acpl_beta_3_dq, acpl_alpha_1_dq, num_pset, z0, z1, y2); (z2, z3) = ACplModule3(acpl_beta_3_dq, acpl_alpha_2_dq, num_pset, z2, z3, y2); (z4, z5) = ACplModule3(-acpl_beta_3_dq, 1, num_pset, z4, z5, y2); z1 *= sqrt(2); z3 *= sqrt(2); z4 *= sqrt(2);
The functions inputSignalModification() and applyTransientDucker() are outlined in clause 5.7.7.4.2 and clause 5.7.7.4.3, respectively. The arrays g1_dq to g6_dq are derived from acpl_gamma1 to acpl_gamma6, and acpl_alpha_1_dq, acpl_beta_1_dq, acpl_alpha_2_dq, acpl_beta_2_dq, acpl_beta_3_dq are derived from acpl_alpha1, acpl_beta1, acpl_alpha2, acpl_beta2, and acpl_beta3 respectively by differential decoding and dequantization as described in clause 5.7.7.7. Note that z5 is used as a temporary variable only and does not constitute an output channel.
The functions ACplModule2(), ACplModule3(), and Transform() are given in the following pseudocode.
Pseudocode 119
Transform(g1, g2, num_pset, x0, x1) { for (sb = 0; sb < num_qmf_subbands; sb++) { for (ts = 0; ts < num_qmf_timeslots; ts++) { interp_g1 = interpolate(g1, num_pset, sb, ts); interp_g2 = interpolate(g2, num_pset, sb, ts); v[ts][sb] = x0[ts][sb]*interp_g1 + x1[ts][sb]*interp_g2; } } } ACplModule2(g1, g2, a, b, num_pset, x0, x1, y) { for (sb = 0; sb < num_qmf_subbands; sb++) { for (ts = 0; ts < num_qmf_timeslots; ts++) { interp_g1 = interpolate(g1, num_pset, sb, ts); interp_g2 = interpolate(g2, num_pset, sb, ts); interp_g1_a = interpolate(g1*a, num_pset, sb, ts); interp_g2_a = interpolate(g2*a, num_pset, sb, ts); interp_b = interpolate(b, num_pset, sb, ts); z0[ts][sb] = 0.5*(x0[ts][sb]*(interp_g1 + interp_g1_a) + x1[ts][sb]*(interp_g2 + interp_g2_a) + y[ts][sb]*interp_b); z1[ts][sb] = 0.5*(x0[ts][sb]*(interp_g1 - interp_g1_a) + x1[ts][sb]*(interp_g2 - interp_g2_a) - y[ts][sb]* interp_b); } } } ACplModule3(b3, a, num_pset, z0, z1, y2) { for (sb = 0; sb < num_qmf_subbands; sb++) {
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)218
for (ts = 0; ts < num_qmf_timeslots; ts++) { interp_b3 = interpolate(b3, num_pset, sb, ts); interp_b3_a = interpolate(b3*a, num_pset, sb, ts); z0[ts][sb] += 0.25*y2[ts][sb]*(interp_b3 + interp_b3_a); z1[ts][sb] += 0.25*y2[ts][sb]*(interp_b3 - interp_b3_a); } } }
In this pseudocode, interpolate() implements the interpolation outlined in clause 5.7.7.3.
5.7.7.6.3 7_X_codec_mode ∈ {ASPX_ACPL_1, ASPX_ACPL_2}
When decoding a 7_X_channel_element, the only codec modes utilizing the A-CPL tool are 7_X_codec_mode = ASPX_ACPL_1 and ASPX_ACPL_2. Here, two parallel Advanced Coupling modules are used, analogously to clause 5.7.7.6.1 for the 5_X_channel_element.
Up to eight input channels are present. Depending on the channel mode and the bitstream element add_ch_base, defined in clause 4.3.3.7.6, the mapping of the six additional channels to A-CPL input and output variables varies, as described in table 201.
Table 201: Input/output channel mapping for 7_X_channel_element
channel_mode 3/4/0.x 5/2/0.x 3/2/2.x add_ch_base n/a 0 1 0 1
x0/z0 Ls L Ls L Ls x1/z2 Rs R Rs R Rs x2/z4 C C C C C x3/z1 Lrs Lw Lw Vhl Vhl x4/z3 Rrs Rw Rw Vhr Vhr x6/z6 L Ls L Ls L x7/z7 R Rs R Rs R
The derivation of the output variables is given in the following pseudocode.
Pseudocode 120
x0in = 2*x0; x1in = 2*x1; u0 = inputSignalModification(x0in); // use decorrelator D0 u1 = inputSignalModification(x1in); // use decorrelator D1 y0 = applyTransientDucker(u0); y1 = applyTransientDucker(u1); if (7_X_codec_mode == ASPX_ACPL_1) { x3in = 2*x3; x4in = 2*x4; (z0, z1) = ACplModule(acpl_alpha_1_dq, acpl_beta_1_dq, num_pset_1, x0in, x3in, y0); (z2, z3) = ACplModule(acpl_alpha_2_dq, acpl_beta_2_dq, num_pset_2, x1in, x4in, y1); } else if (7_X_codec_mode == ASPX_ACPL_2) { (z0, z1) = ACplModule(acpl_alpha_1_dq, acpl_beta_1_dq, num_pset_1, x0in, 0, y0); (z2, z3) = ACplModule(acpl_alpha_2_dq, acpl_beta_2_dq, num_pset_2, x1in, 0, y1); } z1 *= sqrt(2); z3 *= sqrt(2); if (add_ch_base == 1 || channel_mode == 3/4/0.x) { z0 *= sqrt(2); z2 *= sqrt(2); }
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)219
z4 = x2; z6 = x6; z7 = x7; if (add_ch_base == 0) { z6 *= sqrt(2); z7 *= sqrt(2); }
The functions inputSignalModification() and applyTransientDucker() are outlined in clause 5.7.7.4.2 and clause 5.7.7.4.3, respectively, and ACplModule() is defined in clause 5.7.7.5.
The variables num_pset_1 and num_pset_2 indicate the value acpl_num_param_sets of the first and the second acpl_data_1ch() element, respectively.
The arrays acpl_alpha_1_dq and acpl_beta_1_dq are derived from acpl_alpha1 and acpl_beta1 of the first acpl_data_1ch() element and the arrays acpl_alpha_2_dq and acpl_beta_2_dq are derived from acpl_alpha1 and acpl_beta1 of the second acpl_data_1ch() element by differential decoding and dequantization as described in clause 5.7.7.7.
5.7.7.7 Differential decoding and dequantization
To get the quantized values acpl_<SET>_q from the Huffman decoded values acpl_<SET>, where <SET> is an identifier for the A-CPL parameter set type and <SET> ∈ {alpha1, alpha2, beta1, beta2, beta3, gamma1, gamma2, gamma3, gamma4, gamma5, gamma6}, differential decoding shall be done as outlined in the following pseudocode.
Pseudocode 121
// differential decoding for A-CPL // input: array acpl_SET (SET in {alpha1, alpha2, ..., gamma6}) // vector acpl_SET_q_prev // output: array acpl_SET_q num_pset = acpl_num_param_sets; num_bands = acpl_num_param_bands; for (ps = 0; ps < num_pset; ps++) { if (diff_type[ps] == 0) { // DIFF_FREQ acpl_SET_q[ps][0] = acpl_SET[ps][0]; for (i = 1; i < num_bands; i++) { acpl_SET_q[ps][i] = acpl_SET_q[ps][i-1] + acpl_SET[ps][i]; } } else { // DIFF_TIME for (i = 0; i < num_bands; i++) { acpl_SET_q[ps][i] = acpl_SET_q_prev[i] + acpl_SET[ps][i]; } } acpl_SET_q_prev = acpl_SET_q[ps]; }
The quantized values from the last corresponding parameter set of the previous AC-4 frame, acpl_<SET>_q_prev, are needed when delta coding in the time direction over AC-4 frame boundaries.
The dequantized values acpl_alpha_1_dq, acpl_alpha_2_dq, acpl_beta_1_dq, and acpl_beta_2_dq are obtained from acpl_alpha1_q, acpl_alpha2_q, acpl_beta1_q, and acpl_beta2_q using table 202 and table 203 if the quantization mode is set to fine, and using table 204 and table 205 if the quantization mode is set to coarse.
For each parameter band, an index ibeta is obtained from table 202 or table 204 during the dequantization of the alpha values. This value is used in table 203 or table 205, for fine and coarse quantization modes, respectively, to calculate the corresponding dequantized beta value.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)220
Table 202: Alpha values: fine quantization
alpha_q ibeta[alpha_q] alpha_dq[alpha_q] 0 0 -2,000000 1 1 -1,809375 2 2 -1,637500 3 3 -1,484375 4 4 -1,350000 5 5 -1,234375 6 6 -1,137500 7 7 -1,059375 8 8 -1,000000 9 7 -0,940625 10 6 -0,862500 11 5 -0,765625 12 4 -0,650000 13 3 -0,515625 14 2 -0,362500 15 1 -0,190625 16 0 0,000000 17 1 0,190625 18 2 0,362500 19 3 0,515625 20 4 0,650000 21 5 0,765625 22 6 0,862500 23 7 0,940625 24 8 1,000000 25 7 1,059375 26 6 1,137500 27 5 1,234375 28 4 1,350000 29 3 1,484375 30 2 1,637500 31 1 1,809375 32 0 2,000000
Table 203: Beta values: fine quantization
beta_q beta_dq[beta_q] ibeta = 0 ibeta = 1 ibeta = 2 ibeta = 3 ibeta = 4 ibeta = 5 ibeta = 6 ibeta = 7 ibeta = 8
0 0,0000000 0,0000000 0,0000000 0,0000000 0,0000000 0,0000000 0,0000000 0,0000000 0,0000000 1 0,2375000 0,2035449 0,1729297 0,1456543 0,1217188 0,1011230 0,0838672 0,0699512 0,0593750 2 0,5500000 0,4713672 0,4004688 0,3373047 0,2818750 0,2341797 0,1942188 0,1619922 0,1375000 3 0,9375000 0,8034668 0,6826172 0,5749512 0,4804688 0,3991699 0,3310547 0,2761230 0,2343750 4 1,4000000 1,1998440 1,0193750 0,8585938 0,7175000 0,5960938 0,4943750 0,4123438 0,3500000 5 1,9375000 1,6604980 1,4107420 1,1882319 0,9929688 0,8249512 0,6841797 0,5706543 0,4843750 6 2,5500000 2,1854300 1,8567190 1,5638670 1,3068750 1,0857420 0,9004688 0,7510547 0,6375000 7 3,2375000 2,7746389 2,3573050 1,9854980 1,6592190 1,3784670 1,1432420 0,9535449 0,8093750 8 4,0000000 3,4281249 2,9124999 2,4531250 2,0500000 1,7031250 1,4125000 1,1781250 1,0000000
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)221
Table 204: Alpha values: coarse quantization
alpha_q ibeta[alpha_q] alpha_dq[alpha_q] 0 0 -2,000000 1 1 -1,637500 2 2 -1,350000 3 3 -1,137500 4 4 -1,000000 5 3 -0,862500 6 2 -0,650000 7 1 -0,362500 8 0 0,000000 9 1 0,362500 10 2 0,650000 11 3 0,862500 12 4 1,000000 13 3 1,137500 14 2 1,350000 15 1 1,637500 16 0 2,000000
Table 205: Beta values: coarse quantization
beta_q beta_dq[beta_q] ibeta = 0 ibeta = 1 ibeta = 2 ibeta = 3 ibeta = 4
0 0,0000000 0,0000000 0,0000000 0,0000000 0,0000000 1 0,5500000 0,4004688 0,2818750 0,1942188 0,1375000 2 1,4000000 1,0193750 0,7175000 0,4943750 0,3500000 3 2,5500000 1,8567190 1,3068750 0,9004688 0,6375000 4 4,0000000 2,9124999 2,0500000 1,4125000 1,0000000
Table 206: Delta values for dequantizing beta3 values
Quantization mode Delta Fine 0,125 Coarse 0,25
Dequantized acpl_beta_3_dq values are obtained by multiplying the beta3 value by the delta factor corresponding to the quantization mode.
Table 207: Delta values for dequantizing gamma values
Quantization mode Delta Fine 1638/16384 Coarse 3276/16384
Dequantized values g1_dq to g6_dq are obtained by multiplying the gamma value by the delta factor corresponding to the quantization mode.
5.7.8 Dialogue enhancement
5.7.8.1 Introduction
The dialogue enhancement tool is a tool to increase intelligibility of the dialogue in an audio scene encoded in AC-4. The underlying algorithm uses metadata encoded in the bitstream to boost the dialogue in the scene. This technology is especially beneficial for the hearing impaired.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)222
Dialogue enhancement supports enhancement of the dialogue with a user-defined gain. It operates in the QMF domain and is executed before the DRC processing.
Figure 8: Dialogue enhancement in AC-4 decoder
Three different dialogue enhancement modes are available.
Parametric channel independent enhancement In this mode, a parameter subset is transmitted for each channel that contains dialogue. Using these parameters, any dialogue contribution to the respective channels is boosted.
Parametric cross-channel enhancement
In this mode, up to three channels of the content can be used to boost the dialogue in one or more of these channels.
Waveform-parametric hybrid
This mode combines a waveform coded dialogue signal with either parametric enhancement. Both the isolated dialogue signal and the parameter set are transmitted in the bitstream. An additional parameter in the bitstream indicates the balance between waveform and parametric enhancement. Low-complexity decoders may use only the parametric data to perform dialogue enhancement. In the combination with channel independent enhancement, the dialogue signal contains a channel for every transmitted parameter subset, whereas with cross-channel enhancement, a single dialogue signal is transmitted irrespective of the number of transmitted parameter subsets.
Data and control interfaces
Input:
QinDE,[0|1|…] MDE complex QMF matrices of channels containing the regular audio scene.
Qd,[0|1|…]
one or more complex QMF matrices for the channels of an isolated dialogue (waveform-parametric hybrid mode only).
Output:
QoutDE,[0|1|…] MDE complex QMF matrices containing the dialogue enhanced audio scene.
The QinDE, Qd and QoutDE matrices each consist of num_qmf_timeslots columns, and num_qmf_subbands rows.
Control:
GDE Dialogue enhancement gain in dB.
p Parameter vector, consisting of parameter subsets pi corresponding to channel i.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)223
r Rendering vector for spatial positioning of the dialogue.
5.7.8.2 Processed channels
Each of the three decoding methods described in this clause operates, by default, on the QMF domain time-frequency samples of a subset of the three front channels in one AC-4 frame:
mchannel�k,n� ∈ QinDE,channel
for 0 ≤ channel < 3, 0 ≤ k < num_qmf_subbands and 0 ≤ n < num_qmf_timeslots.
As the dialogue may be limited to either, all, or a combination of the three front channels, de_channel_config indicates the channels processed by the dialogue enhancement algorithm, whereas de_nr_channels indicates their absolute number. The two variables are described in clause 4.3.14.3.3.
Consequently, the input time-frequency samples to the dialogue enhancement algorithm are noted as:
mi�k,n� ∈ QinDE,i�de_channel_config{i}��
where de_channel_config{x} is the bit at position x in the binary representation of de_channel_config.
Likewise, output samples are denoted as:
yi�k,n� ∈ QoutDE,i�de_channel_config{i}��
If de_channel_config{x} = 0, the corresponding channel shall not be processed by the dialogue enhancement algorithm and passed through the tool.
5.7.8.3 Dequantization
The parameter index data shall be dequantized using the quantization vectors in table 208 and table 209.
pi'�b,n� = Q���de_par[i][b]�, 0 ≤ � < ��_��_�����
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)224
Table 208: Dequantization vector for channel-independent enhancement mode
Parameter index Parameter value 0 0 1 0,1 2 0,2 3 0,3 4 0,4 5 0,5 6 0,6 7 0,7 8 0,8 9 0,9 10 1,0 11 1,1 12 1,2 13 1,3 14 1,4 15 1,5 16 1,75 17 2,0 18 2,5 19 3,0 20 3,5 21 4,0 22 4,5 23 5,0 24 5,5 25 6,0 26 6,5 27 7,0 28 7,5 29 8,0 30 8,5 31 9,0
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)225
Table 209: Dequantization vector for cross-channel enhancement mode
Parameter index Parameter value -30 -3,0 -29 -2,9 -28 -2,8 -27 -2,7 -26 -2,6 -25 -2,5 -24 -2,4 -23 -2,3 -22 -2,2 -21 -2,1 -20 -2,0 -19 -1,9 -18 -1,8 -17 -1,7 -16 -1,6 -15 -1,5 -14 -1,4 -13 -1,3 -12 -1,2 -11 -1,1 -10 -1,0 -9 -0,9 -8 -0,8 -7 -0,7 -6 -0,6 -5 -0,5 -4 -0,4 -3 -0,3 -2 -0,2 -1 -0,1 0 0 1 0,1 2 0,2 3 0,3 4 0,4 5 0,5 6 0,6 7 0,7 8 0,8 9 0,9 10 1,0 11 1,1 12 1,2 13 1,3 14 1,4 15 1,5 16 1,6 17 1,7 18 1,8 19 1,9 20 2,0 21 2,1 22 2,2 23 2,3 24 2,4 25 2,5 26 2,6 27 2,7 28 2,8 29 2,9 30 3,0
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)226
5.7.8.4 Parameter bands
The parametric enhancement methods rely on segmenting the num_qmf_subbands QMF subbands into de_nr_bands = 8 dialog enhancement parameter bands. The mapping of QMF subbands to parameter bands is described in clause 4.3.14.5.1.
pi'(b, n) (as defined in clause 5.7.8.3) can be mapped to the corresponding QMF band by utilizing table 172.
To improve readability, the time-frequency dependency (b,n) is dropped.
5.7.8.5 Rendering
In modes with cross-channel dialogue enhancement the parametrically generated dialogue-enhancing signal shall be mixed into the audio scene according to the mixing parameters transmitted in the bitstream. When de_nr_channels indicates one involved channel, the dialogue enhancement signal is added to this channel only. That means:
r = �1� For two processed channels, the dequantized mixing coefficient determines the rendering:
r = �r�r�� = � de_mix_coef1�1 − �de_mix_coef1���
For three channels:
r = �r�r�r�� = � de_mix_coef1
de_mix_coef2�max(0.0,(1 − �de_mix_coef1�� − �de_mix_coef2��))�
In the corresponding hybrid mode, the waveform-coded dialogue signal is rendered with the same rendering vector.
5.7.8.6 Interpolation
For a smooth transition between parameter sets, interpolation is applied between successive frames. Clause 5.7.8.7 through clause 5.7.8.9 describe the parameter-based processing for each frame f in terms of a matrix operation.
y = HDE(�) ∙ �mdc�
where m is the subset of channels that dialogue enhancement is applied to (as defined in clause 5.7.8.2) and dc holds the isolated dialogue channels (dc,ch(ts,sb) ∈ Qd,ch).
The application of the dialogue enhancement parameters facilitate changing dialogue enhancement channel configurations between frames by always representing HDE as a 3×6 matrix ĤDE,MC for multichannel and a 2×4 matrix HDE,ST for stereo processing:
HDE,MC = �hmc,00 hmc,01 hmc,02
hmc,10 hmc,11 hmc,12
hmc,20 hmc,21 hmc,22
hmc,03 hmc,04 hmc,05
hmc,13 hmc,14 hmc,15
hmc,23 hmc,24 hmc,25
�
HDE,ST = hst,00 hst,01
hst,10 hst,11
hst,02 hst,03
hst,12 hst,13!
Each of these matrices consists of two square submatrices:
Hsub�,DE,MC = �hmc,�� hmc,�� hmc,02
hmc,10 hmc,11 hmc,12
hmc,20 hmc,21 hmc,22
�
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)227
Hsub2,DE,MC = �hmc,03 hmc,04 hmc,05
hmc,13 hmc,14 hmc,15
hmc,23 hmc,24 hmc,25
�
Hsub1,DE,ST = hst,00 hst,01
hst,10 hst,11!
Hsub2,DE,ST = hst,02 hst,03
hst,12 hst,13!
where the sub1 submatrices indicate the parametric enhancement and the sub2 submatrices indicate the waveform contribution for the waveform-parametric hybrid mode.
For Hsub1,DE,MC (and Hsub1,DE,ST respectively), channels not involved in the enhancement have a 1 on the corresponding diagonal position (for example hmc,11 = 1 if the second channel should not be processed by dialogue enhancement) and 0 in the remaining elements of the corresponding columns and rows (all elements in the second row and second column except hmc,11 respectively).
If dialogue enhancement is run only with parametric data all elements of Hsub2,DE,MC (and Hsub2,DE,ST respectively) are 0.
For the waveform-parametric hybrid mode using channel independent enhancement the rows and columns of Hsub2,DE,MC (and Hsub2,DE,ST respectively) corresponding to channels not involved in dialogue enhancement are 0.
In waveform-parametric hybrid mode using cross-channel enhancement Hsub2,DE,MC (and Hsub2,DE,ST respectively) are 0 in rows corresponding to channels that are not involved in dialogue enhancement. All non-zero components remain in the first column; all other columns are 0.
For example, in a two channel, channel-independent enhancement on the Left and Centre channel of a multichannel signal the dialogue enhancement matrix:
HDE'
= h�� h�� h�� 0
h�� h�� 0 h��!
is represented by its 3×6 transformation:
HDE = �h�� 0 h�� h�� 0 0
0 1 0 0 0 0h�� 0 h�� 0 0 h��
�
Similarly for cross-channel enhancement mode, the dialogue enhancement matrix:
HDE'
= h�� h�� h��h�� h�� h��
!
is represented by:
HDE = �h�� 0 h�� h�� 0 0
0 1 0 0 0 0h�� 0 h�� h�� 0 0
�
In order to obtain the matrix HDE�k,n� for application in QMF time slot n within frame f, an interpolation between the enhancement matrix of the previous frame and the enhancement matrix of the current frame is performed according to:
HDE�k,n� = 1 − n+0.5
N! ∙ HDE�f-1� + n+0.5
N∙ HDE(�)
with N being the frame length in QMF time slots, num_qmf_timeslots.
5.7.8.7 Parametric channel independent enhancement
In this enhancement method, each channel i that has been indicated for processing is processed separately. A set of parameters pi is transmitted for each channel. Additionally, a maximum overall gain Gmax, defined in clause 4.3.14.3.2, limits the user-defined gain g to a range determined at the encoding stage.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)228
Each time-frequency sample mi of each processed channel i is multiplied by the corresponding parameter pi, so that the dialogue-enhanced output signal for this channel can be derived as:
yi = mi + " ∙ pi ∙ mi
where:
g= # 10GDE�� − 1,GDE < Gmax
10Gmax,�� − 1, $%ℎ��&'��
When two channels are processed in a mode with channel independent enhancement parameters �de_method (∈) �0; 2��, an M/S flag (de_ms_proc_flag) is transmitted that indicates whether the parameters are intended for application on the Mid/Side representation of the signal. In that case only one parameter subset is transmitted for application to the Mid signal. The processing equation then becomes:
�y�
y�� = 0,5 ∙ �1 1
1 −1� ∙ 1 + " ∙ p
�0
0 1! ∙ �1 1
1 −1� ∙ �m�
m��
5.7.8.8 Parametric cross-channel enhancement
This method focuses on enhancing the dialogue using a combination of up to three input channels.
The formula for reconstructing the dialogue is noted as follows:
• One processed channel:
y=m+r·g·p ∙�
• Two processed channels:
�y�
y�� = �m�
m��+ �r�
r�� ∙ " ∙ �p� p
�� ∙ �m�
m��
• Three processed channels:
�y�
y�
y�
� = �m�
m�
m�
�+ �r�r�r�� ∙ " ∙ �p� p
�p�� ∙ �m�
m�
m�
�
with:
g= # 10GDE�� − 1,GDE < Gmax
10Gmax,�� − 1, $%ℎ��&'��.
For faster processing, the equation can be written into a single matrix multiplication. Hence, the above equation is represented by:
y = �I + " ∙ r ∙ p�� ∙ m
with I as the identity matrix, r as the rendering vector, and p as the dialogue reconstruction vector; m and y vectors represent the input and output time-frequency samples, respectively, one row per processed channel.
5.7.8.9 Waveform-parametric hybrid
The hybrid dialogue enhancement modes complement the parametric enhancement parameters with transmission of up to three waveforms. The encoder also transmits a trade-off parameter, αc, derived from de_signal_contribution in clause 4.3.14.4.6, that indicates how the enhancement contributions should be divided over the two approaches.
The resulting reconstruction formula for channel independent enhancement can be noted as:
y = I + gp ∙ �'�"�p�! ∙ m + gs ∙ dc
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)229
or
y = (I + gp ∙ �'�"(p) gs ∙ I) ∙ �mdc�.
with diag() returning a square matrix M with the elements of vector p as the diagonal elements, i.e. matrix elements M00 = p0, M11 = p1 and so on.
Similarly, for cross-channel enhancement:
y = (I + gp ∙ r ∙ p r ∙ gs) ∙ �mdc�
with:
gs = * αc ∙ 10 G�� − 1! ," < Gmax
αc ∙ 10Gmax�� − 1! , $%ℎ��&'��
and
gp = * �1 − +c� ∙ 10 G�� − 1! ," < Gmax
�1 − +c� ∙ 10Gmax�� − 1! , $%ℎ��&'��
In the specific case when the M/S flag is set, the equation is:
y = 0,5 ∙ �1 1
1 −1� ∙ 1 + gp ∙ p
�0
0 1! ∙ �1 1
1 −1� 0,5 ∙ gs �11�! ∙ �m
dc�
5.7.9 Dynamic range control tool
5.7.9.1 Introduction
This clause describes a DRC processing unit that is configured by the AC-4 bitstream. For an explanation of how DRC can be used in the scope of an audio codec, refer to document [i.1], clause 6.7.1.1.
DRC is operated in the QMF domain and is executed before QMF synthesis.
Data and control interfaces
Input:
Qin1DRC,[a|b|…] M complex QMF matrices for each of the channels derived from a channel element.
Qin2DRC,[a|b|…] M complex QMF matrices of the same channel that have not been previously processed by the dialogue enhancement tool.
Output:
QoutDRC,[a|b|…] M complex QMF matrices of the DRC processed channels.
The matrices Qin1DRC,ch, Qin2DRC,ch and QoutDRC,ch for each channel ch consist of num_qmf_timeslots columns and num_qmf_subbands rows. Their elements are represented by Qin1ch,k,n, Qin2ch,k,n, and Qoutch,k,n, respectively.
Control:
Lin dialnorm value.
Lout Reference output level for the DRC processor, supplied by the system.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)230
drc_dec_mode_config Up to eight DRC mode configurations that specify DRC processor characteristics.
5.7.9.2 DRC Modes
Compressor operation is controlled by DRC modes. Modes are configured either through a compression curve or by direct gains to be applied, similarly to ETSI TS 102 366 [i.1], clause 6.7.1.
AC-4 supports four default modes:
• home theatre mode,
• flat-panel TV mode,
• portable mode - speakers, and
• portable mode - headphones.
Additional modes may be defined and transmitted in the bitstream.
Each mode is applicable to a range [Lout,min, Lout,max] of output levels. An implementation may provide listener control of choosing a specific DRC mode provided that Lout,min < Lout < Lout,max.
For the default modes, Lout,min and Lout,max are defined in clause 4.3.13.3.1.
For additional modes, Lout,min and Lout,max are transmitted in the bitstream as defined in clause 4.3.13.3.2 and clause 4.3.13.3.3, respectively.
If no applicable mode configuration is available, an implementation may use the last applicable mode configuration transmitted instead.
By default, the decoder shall select the mode with the largest mode ID value for which Lout,min < Lout < Lout,max. For default portable modes, an additional distinction exists between playback over loudspeakers and playback over headphones.
5.7.9.3 Decoding process
5.7.9.3.1 Compression curves
5.7.9.3.1.0 Introduction
When the DRC profile specifies a compression curve (drc_compression_curve_flag == 1), the DRC processor, as depicted in figure 8a, comprises the following functions:
Level computation This function calculates the current signal level.
Gain computation
This function maps the current level into a gain, controlled by a DRC profile. Gain application
This function applies the gain to the QMF samples.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)231
Figure 8a: Dynamic range control processor
5.7.9.3.1.1 Level computation
There are several methods of level computation; an implementation may, for example, employ a level calculation as described in Recommendation ITU-R BS.1770 [i.3], or as a frequency weighted RMS computation. The level shall be expressed relative to the dialnorm. The result of the level calculation is an array Lk,n of levels, one per QMF sample.
5.7.9.3.1.2 Gain computation
The gain computation stage maps the relative level Lk,n into a gain gk,n through the compression curve.
The compression curve consists of seven sections. The total number of sections Ntotal can be derived from the number of boost sections Nboost and the number of cut sections Ncut as:
Ntotal = Nboost + Ncut + 3
The gain value Gk,n should be derived from the level Lk,n and the transmitted compression curve by a piecewise-linear interpolation:
Gk,n =
,---.---/GmaxBoost , Lk,n < LmaxBoost
G�(Lk,n) , LmaxBoost < Lk,n ≤ Lsectionboost
G�(Lk,n) , Lsectionboost < Lk,n ≤ L�,�
0 , L�,� < Lk,n ≤ L�,�� �
G�(Lk,n) , L�,�� � < Lk,n ≤ Lsectioncut
G�(Lk,n) , Lsectioncut < Lk,n ≤ LmaxCut
GmaxCut , Lk,n > LmaxCut
with:
G��x� = Gsectionboost + (GmaxBoost − Gsectionboost) Lsectionboost − x
Lsectionboost − LmaxBoost
G��x� = Gsectionboost L�,� − x
L�,� − Lsectionboost
G��x� = Gsectioncut x − L�,�� �
Lsectioncut − L�,�� �
G��x� = Gsectioncut + �GmaxCut − Gsectioncut� x − Lsectioncut
LmaxCut − Lsectioncut
These gains, expressed in units of dB, should be converted into linear gains:
gk,n = 10Gk,n��
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)232
The gains gk,n should be smoothed with a smoothing factor α:
g~
k,n = α×g~
k,n-1 + �1 0− 0α� × gk,n
where L~
k,n are the smoothed loudness levels using the same smoothing factor α:
L~
k,n = α×L~
k,n-1 + �1 0− 0α� × Lk,n
and:
α=2��
τk,n
where T=64 × 1 000/samplerate the duration of one QMF time slot in ms and
1k,n = � τattack , L~
k,n < Lk,n
τrelease , otherwise
When adaptive smoothing is enabled, i.e. when drc_adaptive_smoothing_flag is set for the active profile, the time constant should additionally be dependent on the speed of level change.
1k,n =
,----.----/ τattack,fast, 10 × log
���Lk,n
L~
k,n
� > drc_attack_threshold
τattack, 0 < 10 × log��
�Lk,n
L~
k,n
� ≤ drc_attack_threshold
τrelease, 0 ≤ 10 × log��
�L~
k,n
Lk,n� ≤ drc_release_threshold
τrelease,fast , 10 × log��
�L~
k,n
Lk,n� > drc_release_threshold
NOTE: See clause 4.3.13.4.12 through clause 4.3.13.4.15 for the definition of τattack and τrelease.
A reset flag (drc_reset_flag) is transmitted for profiles with compression curves to indicate when a reset of DRC processing is needed. When drc_reset_flag is set, the state of the smoothing operation should be reset by:
g~
k,n = gk,n
L~
k,n = Lk,n
5.7.9.3.2 Directly transmitted DRC gains
For profiles where DRC gains are transmitted (drc_compression_curve_flag == 0), there are four configurations supported for transmitting gain values in the AC-4 bitstream:
• single wideband gain for all channels,
• wideband gain for each channel group,
• multiband for each channel group, 2 bands, and
• multiband for each channel group, 4 bands.
The channel-dependent gains correspond to channel groups. For a 5_X_channel_element, for example, there are three channel groups chg [Lf, Rf, LFE], [Cf], [Ls, Rs]. The update rate of DRC gains depends on the frame length.
NOTE: The definition of the DRC gain bands, channel groups, and update rate is given in clause 4.3.13.3.8, clause 4.3.13.7.1, and clause 4.3.13.7.2, respectively.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)233
After differential decoding of the gains, resulting in drc_gain[chg][sf][band] per channel group, the gains for each channel ch are adjusted to reflect dB2 values.
gk,n,ch = drc_gain2chg32sf32band3|k�band,�∈sf,��∈chg
5.7.9.3.3 Application of gain values
As a last step, the DRC tool shall map the dialnorm level to the output level Lout, and apply the gains that have either been calculated per clause 5.7.9.3.1.2 or transmitted as per clause 5.7.9.3.2.
Qoutch,k,n = 2Lout�Lin
� ∙ gk,n,ch ∙ Qin1ch,k,n
5.7.9.4 Transcoding to a AC-3 or E-AC-3 format
When an AC-4 bitstream is transcoded into a format such as AC-3 or E-AC-3, no DRC processing should be applied. Instead, the encoder or transcoder should be configured with the compression curve indicated by the field drc_eac3_transcode_curve in table 159.
6 Decoding the AC-4 bitstream
6.1 Introduction Clause 4 specifies the details of the AC-4 bitstream syntax and clause 5 specifies the algorithmic details of the tools used within AC-4. This clause describes the complete AC-4 decoding process and by this makes a connection between the bitstream elements and the AC-4 tools. Figure 9 shows a flow diagram of the AC-4 decoding process for one substream.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)234
Figure 9: Flow diagram of the decoding process (one substream)
6.2 Decoding process
6.2.1 Input bitstream
The input bitstream typically comes from a transmission or storage system. To decode an AC-4 bitstream, all raw_ac4_frame elements shall be extracted according to the used transport format and presented to the AC-4 decoder.
6.2.2 Structure of the bitstream
6.2.2.0 Overview
An AC-4 bitstream is roughly structured in several containers, namely:
Raw AC-4 frame The actual codec frame, consisting of a table of contents plus several byte aligned substreams.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)235
table of contents (TOC) Signalling information on how to treat the content of the substream container of the AC-4 frame.
Substream
Encoded audio and metadata, accompanied by a metadata skip area and a coding paradigm. Presentation information
Details on the presentations available in the AC-4 bitstream. Substream information
Details on the substreams available in the AC-4 bitstream. Metadata
Audio metadata associated with the audio data encoded in the AC-4 frame.
Details on the information contained in these structures can be found in the following clauses.
Figure 9a: AC-4 container structure
6.2.2.1 Raw AC-4 frame
Each raw AC-4 frame contains a table of contents (TOC) and at least one byte-aligned substream. The TOC can be considered as the bitstream inventory where all information that is important for the overlaying system resides. Substreams are decodable units that represent a specific channel configuration (mono, stereo, 3.0, 5.1, 7.1).
In transport scenarios, a raw AC-4 frame is typically embedded in a transport container, while it would be considered an AC-4 sample in ISO base media file formats. Details on embedding an AC-4 frame into ISO base media files can be found in annex E.
The syntax of the raw_ac4_frame() element is described in clause 4.3.1.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)236
6.2.2.2 Table of contents
The table of contents (TOC) contains the TOC data and at least one presentation as well as a substream index table. As the bitstream inventory the TOC provides information about and access to individual substreams.
The TOC data contains stream specific metadata, such as:
• bitstream version, which signals if the bitstream is decodable by a decoder compliant to this AC-4 specification;
• sequence counter, which can be used to detect splices in the bitstream;
• information about buffer model such as current average bit rate and bit-rate smoothing buffer level;
• information about I-Frames usable as random access points;
• base sampling frequency, which serves as a common denominator for the output sampling frequency of the audio;
• total number of presentations.
Additionally, the TOC contains the presentation information as well as a substream index. The index provides easy access to each of the substreams by specifying the absolute number and the size of each of the substreams. Further details about the presentation information can be found in clause 6.2.2.3.
Because of the fact that all the above information is in the TOC, it is not necessary to parse the complete bitstream to decode a single presentation, given that more presentations are available.
The TOC is contained in the bitstream element ac4_toc(), which is described in clause 4.3.3.2.
6.2.2.3 Presentation information
Typically, a presentation in AC-4 describes a set of substreams that is to be decoded and presented simultaneously. Mixing of decoded content only occurs within one presentation. That means only one presentation has to be decoded at a time. On the other hand, one substream can be part of several presentations.
Examples for presentations are:
Single substream The common case: a single multichannel stream in one presentation.
Multi-Language
A set of 'music and effects' combined with multiple dialogue language tracks. Main/Associate
A set of main and associate signal, the latter of which could contain scene description or the director's comments.
Dialogue enhancement track
A separate track used for boosting the dialogue (see clause 5.7.8 for details).
The presentation information container tells the decoder which substreams to decode, where to find them, and how they shall be presented. The ac4_presentation_info element is described in clause 4.3.3.3.
6.2.2.4 Substream
A substream contains the actual audio data, as well as the corresponding metadata.
Additionally, the size of the audio data is given in the header of the substream, so that the metadata can be accessed without the need of parsing the audio data.
The ac4_substream() element is specified in clause 4.3.4.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)237
6.2.3 Selecting and decoding a presentation
To successfully select and decode an appropriate presentation, a decoder system has to execute the following steps:
1) Create a list of presentations available in the bitstream.
Initially, the number of presentations, n_presentations, has to be derived from the bitstream elements b_single_presentation and b_more_presentations in ac4_toc(), as described in clause 4.3.3.2.
For each of these n_presentations, parse the ac4_presentation_info() fields. Here, single substream presentations as well as multi-substream presentations described by the presentation_config element can be found, as indicated in clause 4.3.3.3.3.
2) Select the presentation that is appropriate for the decoder and the output scenario.
Information available to support this selection is mainly language type, availability of associated audio, or type of audio (multichannel for speaker rendering, or a pre-virtualized rendition for headphones).
A decoding system should employ a user agent to decide without the need for direct user interaction.
NOTE 1: The number and the order of presentations in the stream may change.
3) Select the substreams associated with the presentation.
From each of the ac4_substream_info() fields associated with the presentation_config, extract the presentation-specific metadata such as channel mode, bit-rate information, content type, and the substream index, according to clause 4.3.3.7.
The substream_index field indicates the substream associated with the presentation. It can be used as an index to the substream_index_table.
For each of the substreams, the substream_index_table() provides the total size of the substream. If that is the case (b_size_present is true), the offset to the substream data relative to the end of the ac4_toc() element in the elementary stream can be calculated as follows.
Pseudocode 122
n = substream_index; substream_n_offset = payload_base; for (s = 0; s < n; s++) { substream_n_offset += substream_size[s]; }
NOTE 2: payload_base is defined in clause 4.3.3.2.11.
4) Decode the selected substream(s).
Instructions on how to decode the selected substream(s) can be found in clause 6.2.5.0 and the following clauses.
5) Mix the decoded substream(s).
Instructions on how to mix the substream(s) can be found in clause 6.2.16.0.
6) Render the mixed audio.
Instructions on how to render the audio can be found in clause 6.2.17.0.
6.2.4 Buffer model
Decoders implemented in accordance with the present document shall provide a minimum input buffer of size N bytes where N is set according to:
N=v× RstreamRframe5
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)238
Rstream and Rframe are the total bit rate and the frame_rate, respectively, and v= 6 6, RFrame ≤ 60
12, else.
See table 210 for typical values of N.
NOTE 1: VBR streams (see clause 4.3.3.2.4) may contain raw_ac4_frames exceeding N.
NOTE 2: Interpretation of total bit rate and input buffer are subject to AC-4 applications.
Table 210: Minimum buffer size N in bytes for typical frame rates and bit rates at sample rate of 48 kHz
Bit rate [bps] frame_rate 25 30 500 60 120
48 000 1 440 1 200 720 600 600 96 000 2 880 2 400 1 440 1 200 1 200 160 000 4 800 4 000 2 400 2 000 2 000 1 536 000 46 080 38 400 23 040 19 200 19 200
6.2.5 Decoding of a substream
6.2.5.0 Substream types
As described in clause 4.2.4, three different types of AC-4 substreams exist:
• AC-4 substream,
• AC-4 high sampling frequency extension substream, and
• EMDF payloads substream.
6.2.5.1 Decoding of an AC-4 substream
A decoder shall decode audio_data() and metadata() in the AC-4 substream. Both bitstream elements can be accessed independently of each other by using the variable audio_size.
Decoding of the channel elements within audio_data() is described in clause 6.2.6.0 and the following clauses.
The bitstream elements contained in metadata() shall be decoded from the bitstream using the information provided in clause 4.2.14 and clause 4.3.12 to clause 4.3.15.0. The decoded information can be found in the following clauses:
• for rendering in clause 6.2.17.0;
• for dialogue enhancement in clause 6.2.12; and
• for dynamic range control in clause 6.2.13.
6.2.5.2 Decoding of an AC-4 HSF extension substream
Decoding of the HSF substream can be achieved using the information derived from clause 5.4. Afterwards, the SAP tool and the IMDCT described in clause 5.3 and clause 5.5 need to be employed. No QMF domain processing is required.
6.2.5.3 Decoding of an EMDF payloads substream
A decoder shall be able to skip an extensible metadata delivery format (EMDF) substream, and or may ignore its contents.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)239
6.2.6 Spectral frontend decoding
6.2.6.0 audio_data element types
The channel element contained in the audio_data element of an AC-4 substream can be one of four types, signalled by the channel_mode, as indicated in clause 4.2.6.
Within the channel elements, audio tracks are encoded as sf_data elements. These elements are associated with an sf_info element, as well as with a spectral frontend type.
Tracks with an associated spectral frontend type of ASF (spec_frontend = 0) shall be processed by the audio spectral front end for the decoding of spectral lines, as described in clause 6.2.6.4. Tracks associated with the type SSF (spec_frontend = 1) shall be processed by the SSF tool, as described in clause 6.2.6.5. The output of both spectral frontends is a vector of spectral lines for each track, as well as associated information about the subsequent windowing.
6.2.6.1 Mono decoding
If channel_mode is mono, the bitstream contains only one track stored as an sf_data element within a mono_data element. The decoding of the sf_data element depends on the spectral frontend type and on the sf_info element and is done in the following steps:
1) Parsing of sf_info which includes transform and psy information for the spectral frontend type.
2) Parsing of spectral frontend data depending on spectral frontend type.
3) Decoding of spectral frontend data depending on spectral frontend type.
The resulting mono track shall be considered the Centre channel (C), and shall be directly routed to the IMDCT stage.
6.2.6.2 Stereo decoding
If channel_mode is stereo, the bitstream contains up to two audio tracks stored as sf_data elements within a channel_pair_element. Similarly to the mono decoding, the decoding of the sf_data element is done in steps described in clause 6.2.6.1. If the b_enable_mdct_stereo_proc is true, a common sf_info element is present that shall be used for the decoding of both sf_data elements.
Subsequently, the decoded audio tracks are routed into the stereo and multichannel processing tool, described in clause 5.3. Depending on the stereo_codec_mode, the tool assigns dedicated speaker locations to the tracks, and by that turning them into dedicated channels. The spectral lines of these channels shall subsequently be processed by the IMDCT tool, as described in clause 6.2.7.
6.2.6.3 Multichannel audio decoding
Similarly to stereo decoding, first, all the sf_data elements are decoded, followed by routing the decoded tracks into the stereo and multichannel processing tool. This tool returns dedicated audio channels. The spectral lines of these channels shall subsequently be processed by the IMDCT tool.
If channel mode is 5.1 or 7.1, the audio track acquired from the first sf_data element shall be mapped to the LFE channel, and shall be directly routed to the IMDCT stage, i.e. is not passed into the stereo and multichannel processing tool.
6.2.6.4 Audio spectral front end
The asf_section_data and asf_scalefac_data elements of the sf_data element are parsed and helper elements are set up. See clause 4.3.6.3 and clause 4.3.6.5. The asf_spectral_data is Huffman decoded using the helper elements. See clause 5.1.2 for details of the Huffman decoding process for the quantized spectral lines. Next, the quantized spectral lines shall be reconstructed and scaled as described in clause 5.1.3. If the scaled spectral lines do not belong to a full block, a reordering of the spectral lines shall be performed as described in clause 5.1.5.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)240
6.2.6.5 Speech spectral frontend (SSF)
The sf_data element contains just the ssf_data element if the spec_frontend value is '1'. Either one or two ssf_granule elements are present in the ssf_data element. Each ssf_granule element is decoded into a vector of spectral lines as specified in clause 5.2.
6.2.7 (IMDCT) and window overlap/add
The decoding steps described in the preceding clauses result in a set of frame_length (reordered) spectral lines for each spectral frontend channel. The inverse MDCT transform converts the blocks of spectral lines into blocks of time samples.
The individual blocks of time samples shall be windowed, and adjacent blocks shall be overlapped and added together in order to reconstruct a continuous time PCM audio signal for each spectral frontend channel. The window and overlap/add steps are described along with the inverse MDCT transform in clause 5.5.
6.2.8 QMF analysis
The time domain samples of each channel are transformed individually into the QMF domain by the use of a QMF analysis filter as described in clause 5.7.3. Depending on the frame_length, 32, 30, 24, 15, 12 or 6 QMF slots are the result of the QMF analysis. Each QMF slot is the result of one QMF filtering step with num_qmf_subbands time domain samples as input. The QMF analysis step is needed for the tools which operate in the QMF domain.
6.2.9 Companding
The companding tool mitigates pre- and post-echo artefacts. The encoder applies attenuation to signal parts with higher energy, and gain to signal parts with lower energy. The decoder applies the inverse process, which effectively shapes the coding noise by the signal energy.
Whether or not a channel is processed by the companding tool is dependent on the channel element and the codec mode. Table 211 lists the channels that shall be processed by the companding tool. Channels that are not listed shall not be processed.
Table 211: Channels processed by the companding tool
Channel element Codec mode Channels in companding single_channel_element 1 C channel_pair_element 1 L, R channel_pair_element 2,3 L 3_0_channel_element 1 L, R, C 5_X_channel_element 1 L, R, C, Ls, Rs 5_X_channel_element 2,3 L, R, C 5_X_channel_element 4 L, R 7_X_channel_element 2,3 L, R, C, Ls, Rs
The companding tool is described in clause 5.7.5.
Table 211 also specifies the order of the channels in the companding_control() element.
6.2.10 Advanced spectral extension - A-SPX
Spectral extension includes translating the spectral content of suitable low-frequency regions of the signal to higher frequencies and adjusting them such that they perceptually match the high-frequency content that is present in the original signal. Waveform coding may be employed in addition, either when critical signal frequency components are not present in the low-frequency regions, or to reproduce transients more accurately than with spectral extension alone.
An AC-4 decoder shall process the QMF domain channels except for the LFE channel with the A-SPX tool after the companding processing has been applied. A-SPX shall be active for all codec modes except for the SIMPLE codec mode (0).
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)241
Table 212 lists the input A-SPX channels for each channel element and codec mode. Individually processed channels (as QinASPX,a) are denoted as single letters, channels processed simultaneously (as QinASPX,a and QinASPX,b) are combined by surrounding parenthesis. The order of the listing indicates the order of corresponding aspx_data elements in the bitstream, aspx_data_1ch for individually processed channels, aspx_data_2ch for simultaneously processed channels.
Table 212: Channels processed by A-SPX tool
Channel element Codec mode Channel mode Channels in A-SPX single_channel_element 1 - C channel_pair_element 1 - (L, R) channel_pair_element 2,3 - L 3_0_channel_element 1 - (L, R),C 5_X_channel_element 1 - (L, R), (Ls, Rs), C 5_X_channel_element 2,3 - (L, R), C 5_X_channel_element 4 - (L, R) 7_X_channel_element 1 3/4/0.x (L, R), (Ls, Rs), C, (Lrs, Rrs) 7_X_channel_element 1 5/2/0.x (L, R), (Lw, Rw), C, (Ls, Rs) 7_X_channel_element 1 3/2/2.x (L, R), (Ls, Rs), C, (Lth, Rth) 7_X_channel_element 2,3 - (L, R), (Ls, Rs), C
The A-SPX tool is described in clause 5.7.6.
NOTE: LSx and RSx are not original channels but already processed (downmixed) signals passed into the A-SPX module pre-processed by another module.
6.2.11 Advanced coupling - A-CPL
Advanced coupling reconstructs multichannel content out of a lower number of input channels, using encoder controlled side information. As a special case, a stereo signal can be generated from a mono downmix.
An AC-4 decoder shall process the QMF domain channels with the A-CPL after the A-SPX processing has been applied. A-CPL shall be active for all codec modes except for the SIMPLE and ASPX codec modes (0,1). Table 213 indicates which channels shall be processed by the A-CPL tool.
Table 213: Channels processed by A-CPL tool
Channel element Codec mode Channels in A-CPL channel_pair_element 2, 3 L, R 5_X_channel_element 2, 3, 4 L, R, Ls, Rs, C 7_X_channel_element 2, 3 All channels except LFE
The A-CPL tool is described in clause 5.7.7.
6.2.12 Dialogue enhancement
Dialogue enhancement control metadata can be specified in the bitstream. When the decoder is configured to enable dialogue enhancement, the dialogue enhancement tool uses this metadata to apply a gain on the dialogue in the signal according to a system-defined gain.
Table 214 lists the input channels for the dialogue enhancement tool.
Table 214: Channels processed by the dialogue enhancement tool
Channel element Channels in dialogue enhancement single_channel_element C channel_pair_element L,R 3_0_channel_element L,R,C 5_X_channel_element L,R,C 7_X_channel_element L,R,C
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)242
The dialogue enhancement tool is described in clause 5.7.8.
6.2.13 Dynamic range control
Dynamic range control metadata can be specified in the bitstream. The DRC tool uses this metadata to apply gains to the channels in the QMF domain in order to adjust the dynamic range as specified in clause 5.7.9.
The DRC tool takes two inputs per channel: the audio signal which is output by a preceding tool (usually dialogue enhancement) and the signal that is used to measure levels and drive the side chain. The signal driving the side chain shall be identical to the input of dialogue enhancement (i.e. it does not include the dialogue enhancements).
6.2.14 QMF synthesis
A QMF synthesis filter shall transform each audio channel from the QMF domain back to the time domain as specified in clause 5.7.4.
6.2.15 Sampling rate converter
For all values of the frame_rate_index, the decoder may be operated at external sampling frequencies of 48 kHz, 96 kHz, or 192 kHz. For frame_rate_index = 13, 44,1 kHz is also a permissible external sampling frequency.
To adjust the sampling frequency to the external sampling frequency, a sampling rate converter shall be used. The decoder resampling ratio shall be as specified in clause 4.3.3.2.6.
The sampling rate converter should use high-quality anti-aliasing filters.
6.2.16 Mixing substream outputs
6.2.16.0 Introduction
This clause describes how to mix several substreams contained within one presentation. This includes:
• mixing main + associated audio; and
• mixing music and effects (M+E) + dialogue.
Dialogue enhancement is not part of this clause.
The decoding of a presentation that contains multiple substreams, as indicated by a b_single_substream value of false, and where presentation_config has a value of 0, 2, 3, or 4 shall involve a mixing of the substream outputs as specified in this clause. The channel configuration of the mixer output matches the channel configuration of the main or the music and effects (M+E) substream. No additional channels are allowed for the dialogue or associate substream, except for a mono channel.
The mixing metadata for main/associate mixing and for music and effects (M+E)/dialogue mixing are not necessarily sent with each frame, and not necessarily with each I-frame, although this is encouraged. The mixing metadata values remain valid until new ones are transmitted in the same stream or until a splice is detected.
Before mixing the substreams, they need to be at the same reference level. If levelling has not already been done in DRC processing as described in clause 5.7.9.3.3, the decoder shall apply a gain to the substream. This levelling before mixing shall not be done if both the presentation_version is 0 and the dialnorm values of the music and effects (M+E) and dialogue substreams differ. If levelling is to be done, it shall be done as follows:
poutch(n) = 10Lout��������
�� ∙ pinch(�) Here, pinch and poutch refer to the input and output samples of each channel ch, respectively. Lout is the reference output level, supplied by the system.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)243
6.2.16.1 Mixing music and effects (M+E) and dialogue
While mixing the dialogue tracks, a single user-agent provided gain g_dialog ∈ [-∞, g_dialog_max] dB should be applied to all dialogue tracks. This allows for a user-controlled adjustment of the relative dialogue level. The user agent default value for g_dialog shall be 0 dB.
Each track that is only present in the music and effects (M+E) substream output is left untouched. All other tracks are mixed with the corresponding track from the dialogue substream output:
Y_mixi = X_music_effectsi + X_dialogi × pan_dialog_gain[ch] × g_dialog,
where Y_mixi is a mixer output sample and X_music_effectsi and X_dialogi are the corresponding mixer input samples from the music and effects (M+E) track and the dialogue track involved in the mixing process.
The pan_dialog_gain[ch] array is a gain factor per track that shall be derived from the bitstream element pan_dialog, which indicates the panning of the dialogue given as angle in degree. The calculation of the pan_dialog_gain[ch] factor from pan_dialog shall be done based on the mandatory requirements for the gain factor which are shown in table 214a. For mono and stereo dialogue streams, Y_mix works as usual, i.e. there is one pan_dialog element per track. For dialogue streams with more than two tracks, two pan_dialog elements are present. The dialogue tracks to which these calculations apply are signalled by the value of pan_signal_selector. Tracks that do not have a pan value are not to played back.
Table 214a: pan_dialog_gain[ch]
pan_dialog in degree pan_dialog_gain[L] pan_dialog_gain[C] pan_dialog_gain[R] 330 1,0 0,0 0,0 0 (in case of output is 5.1) 0,0 1,0 0,0 0 (in case of output is 2.0) 0,5 n/a 0,5 30 0,0 0,0 1,0
6.2.16.2 Mixing main and associate
While mixing the associate channels, a single user-agent provided gain g_assoc [-∞, 0] dB should be applied to all associate channels. This allows for a user-controlled adjustment of the relative associate level. The user agent default value for g_assoc shall be 0 dB.
The extended_metadata part of the associate substream shall provide gain values for a gain-adjustment of the channels in the main substream if an adjustment needs to be done. If no gain values are given, a default of 0 dB shall be used. First, the gain for the Centre channel of the main substream, if available, is adjusted using scale_main_center. Next, the gain for the two front channels L and R of the main substream is adjusted using scale_main_front. And finally, the gain for all channels of the main substream is adjusted using scale_main. If required by the use case, the scale factors scale_main_center, scale_main_front and scale_main may be modified before applying them.
If the channel_mode of the associate substream is mono, a pan_associated value is specified in the extended_metadata. The panning value in degrees (°), or a default of 0°, shall be used to pan the mono signal to all channels available in the main substream. This is achieved by converting the panning value into an array of panning gains, one entry for each channel available in the main substream. The mixer output then is given by:
Y_mixi,ch = X_main_adji,ch + X_associatei,mono × 7��_"�'�[8ℎ] × "_���$8, where Y_mixi,ch is a mixer output sample for channel ch and X_main_adji,ch is the corresponding mixer input sample for channel ch from the gain-adjusted signal in the main channel. X_associatei,mono is the corresponding mixer input sample from the mono associate channel and pan_gain[ch] is the panning gain for channel ch.
If the channel_mode of the associate substream is not mono, all channels that are only present in the main substream output are replaced by the gain-adjusted signal. Channels available in both, the main and associate substream, are mixed using the gain-adjusted main signal:
Y_mixi = X_main_adji + X_associatei,
where Y_mixi is a mixer output sample and X_main_adji and X_associatei are the corresponding mixer input samples from the gain-adjusted main channel signal and the associate channel involved in the mixing process.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)244
6.2.16.3 Mixing music and effects (M+E), dialogue and associate
The mixing of music and effects (M+E), dialogue and associate substream outputs is done in two stages. First, the M+E and the dialogue channels are mixed as specified in clause 6.2.16.1. The resulting main signal is mixed with the associate signal as described in clause 6.2.16.2.
6.2.17 Rendering a presentation
6.2.17.0 Introduction
When the number of loudspeakers does not match the number of encoded audio channels either downmixing or upmixing is required to render the complete audio programme. It is important that the equations for downmixing and upmixing be standardized, so that programme providers can be confident of how their programme is rendered over systems with a varying number of loudspeakers. Using standardized rendering equations programme producers can monitor how the rendered version will sound and make any alterations necessary to achieve acceptable results for all listeners.
The source channel modes for downmixing are 7.X, 5.X, 3.0, and stereo. The destination channel modes for downmixing are 5.X, stereo, and mono.
The source channel modes for upmixing are mono, stereo, 3.0, and 5.X. The destination channel modes for upmixing are stereo, 3.0, 5.X, and 7.X.
The downmix and upmix operations are cascaded. For example, to downmix from seven channels to two channels, the seven channels are first downmixed to five channels and then five channels to two channels. Similarly, to upmix from one channel to five channels, the mono channel is upmixed to two channels, two channels to three channels, and three channels to five channels.
The mixing metadata are not necessarily sent with each frame, and not necessarily with each I-frame, although this is encouraged. The mixing metadata values remain valid until new ones are transmitted in the same stream or until a splice is detected.
6.2.17.1 Generalized rendering matrix and equation
The generalized rendering matrix, R, is as follows:
9 r�,� r�,� ⋯ r�,���r�,� r�,� ⋯ r�,���⋮ ⋮ ⋱ ⋮
rn-�,� rn-�,� ⋯ rn-1,n-1
: where n is the number of channels listed in table D.1.
The elements r 0,0 to r n-1,n-1 are defined for each combination of available input and output channels.
The rendering equation is as follows:
;<<= y_out
�
y_out�
⋮
y_outn-1>??@ = 9 r�,� r�,� ⋯ r�,���
r�,� r�,� ⋯ r�,���⋮ ⋮ ⋱ ⋮
rn-�,� rn-�,� ⋯ rn-1,n-1
: ;<<= x_in
�
x_in�
⋮
x_inn-1>??@
where x_in is a matrix of sample values of input channels, and y_out is a matrix of sample values for output channels.
6.2.17.2 Downmixing from two channels into one channel
The rendering matrix R2:1, used to downmix two channels (L,R) into one channel (C), has the elements r1,0 and r1,2 set to 1 and all other elements of the matrix set to 0.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)245
6.2.17.3 Downmixing from three channels into two channels
The rendering matrix R3:2 used to downmix from three channels into two channels is dependent on the bitstream element preferred_dmx_method, described in clause 4.3.12.2.19.
The non-zero entries of the rendering matrix R3:2 are given in table 215.
Table 215: Matrix entries for 3- to 2-channel downmix matrix
preferred_dmx_method r0,0r2,2 r0,1r2,1 1 1 loro_cmg 2 1 ltrt_cmg 3 1 ltrt_cmg
The loro_cmg variable is derived from the bitstream element loro_center_mixgain, as described in clause 4.3.12.2.8, and ltrt_cmg is derived from the bitstream element ltrt_center_mixgain, described in clause 4.3.12.2.13.
All other elements of the rendering matrix are set to 0.
Additionally, loro_corr_gain (preferred_dmx_method = 1) or ltrt_corr_gain (preferred_dmx_method = 2,3) shall be applied to the downmixed signal if available in the bitstream, as described in clause 4.3.12.2.11 and clause 4.3.12.2.16.
6.2.17.4 Downmixing from five channels into two channels
The rendering matrix R5:2, used to downmix from five channels into two channels, is dependent on the bitstream element preferred_dmx_method, described in clause 4.3.12.2.19.
The non-zero entries of the rendering matrix R5:2 are given in table 216.
Table 216: Matrix entries for 5- to 2-channel downmix matrix
preferred_dmx_method r0,0r2,2 r0,1r2,1 r0,3 r2,4 r0,4 r2,3 r0,18 r2,18
1 1 loro_cmg loro_smg loro_smg 0 0 lfe_mg 2 1 ltrt_cmg -ltrt_smg ltrt_smg -ltrt_smg ltrt_smg lfe_mg 3 1 ltrt_cmg -ltrt_smg ×
(+1,8 dB) ltrt_smg × (+1,8 dB)
-ltrt_smg × (-3,2 dB)
ltrt_smg × (-3,2 dB)
lfe_mg
The loro_cmg and loro_smg variables are derived from the bitstream elements loro_center_mixgain and loro_surround_mixgain, respectively, as described in clause 4.3.12.2.8. Likewise, ltrt_cmg and ltrt_smg are derived from the bitstream elements ltrt_center_mixgain and ltrt_surround_mixgain, respectively, described in clause 4.3.12.2.13. lfe_mg is derived from the bitstream element lfe_mixgain as described in clause 4.3.12.2.18.
All other elements of the rendering matrix are set to 0.
Additionally, loro_corr_gain (preferred_dmx_method = 1) or ltrt_corr_gain (preferred_dmx_method = 2,3) shall be applied to the downmixed signal if available in the bitstream, as described in clause 4.3.12.2.11 and clause 4.3.12.2.16.
6.2.17.5 Downmixing from seven channels into five channels
The rendering matrix R7:5 , used to downmix seven channels into five channels, is dependent on the channel mode, as well as on the bitstream element add_ch_base.
Centre channel (r 1,1 = 1) and LFE (if present) are routed through (r 18,18 = 1).
The remaining non-zero entries of the rendering matrix R7:5 are given in table 217.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)246
Table 217: Additional matrix entries for 7-channel to 5-channel downmix matrix
channel mode add_ch_base r0,0 r2,2 r0,13r2,14 r0,15r2,16 r3,3r4,4 r3,7r4,8 r3,13 r4,14 r3,15r4,16 3/4/0.x n/a 1 0 0 0,707 0,707 0 0 5/2/0.x 0 1 0,707 0 1 0 0 0
1 1 0 0 0,707 0 0,707 0 3/2/2.x 0 1 0 0,707 1 0 0 0
1 1 0 0 0,707 0 0 0,707
6.2.17.6 Upmixing from one channel into two channels
The rendering matrix R1:2, used to upmix from a mono channel to stereo, has the elements r1,0 and r1,2 set to 0,707 and all other elements of the matrix set to 0.
6.2.17.7 Upmixing from two channels into three channels
The rendering matrix R2:3, used to upmix from two channels to three channels, has the elements r0,0 and r2,2 set to 1 and all other elements of the matrix set to 0.
6.2.17.8 Upmixing from three channels into five channels
The rendering matrix R3:5, used to upmix from three channels to five channels, has the elements r0,0, r1,1 and r2,2 set to 1 and all other elements of the matrix set to 0.
6.2.17.9 Upmixing from five channels into seven channels
The rendering matrix R5:7, used to upmix from five channels into seven channels, has the elements r0,0, r1,1, r2,2, r3,3 and r4,4 set to 1. The LFE channel (if present) is routed through (r18,18 = 1) as well. All other elements of the matrix are set to 0.
6.2.18 Decoding audio in sync with video
In order to achieve A/V sync, a synchronization mechanism, usually employing presentation timestamps, needs to be defined in application standards referencing the present document. If such a timestamp is defined and applies to a certain audio frame, it should indicate the presentation time of the sample at the reference point at sample position frame_length/2 of the composition buffer.
6.2.19 Switching streams while decoding
AC-4 has been designed to allow for a seamless switching of AC-4 streams during the run time of an AC-4 decoder. When the switch happens at an I-frame boundary, which is the case when the first frame of the new stream is an I-frame, the AC-4 decoder shall produce a flawless output. When the switch does not happen at an I-frame boundary, the AC-4 decoder shall restore the audio output before or with the reception of the next I-frame at the latest. One aspect that allows for a seamless switching is the alignment of the frame signal with the control signal as described in clause 5.6. The seamless switching capability of an AC-4 decoder facilitates several use cases.
In an adaptive streaming application, the bit rate can be seamlessly adjusted according to the available transmission channel bandwidth. The server, which for example can be an MPEG-DASH server (see ISO/IEC 23009-1 [1]), just needs to make sure that the AC-4 stream to be switched to contains an I-frame at the switching point. This information is easily available to the server without much parsing effort since the b_iframe_global flag is part of ac4_toc(), the first element of a raw_ac4_frame() structure. When the switch is done as described on an I-frame boundary, the AC-4 decoder can produce a flawless output signal.
When AC-4 is used in a TV application, a user initiated switch of streams can be detected by evaluating the sequence_counter as described in clause 4.3.3.2.2. No external signalling of a stream switch to the AC-4 decoder is needed. Since the switch might occur at a point in time when no I-frame is present in the new stream, it is possible that the audio output signal has a reduced coverage for a short period of time until the next I-frame is received and the full output signal is available again.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)247
A switch of streams at the broadcast side, often referred to as splicing, can be either a controlled or uncontrolled splice. A controlled splice is a splice that happens on an I-frame boundary and should be done similarly to the adaptive streaming use case described above. In addition to selecting an I-frame as the first AC-4 frame after the splice point, the sequence_counter of this AC-4 frame should be set to 0 to indicate the splice. If all this is done, the AC-4 decoder can produce a seamless output signal. If a controlled splice is not possible, the sequence_counter of the first AC-4 frame of the new stream should be set to 0 to indicate the splice to the decoder. An AC-4 decoder shall use this splice indication to ignore any information from previous frames when decoding the first frame after a splice.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)248
Annex A (normative): Huffman codebook tables
A.0 Introduction All Huffman codebook tables are available in the file ts_103190_tables.c contained in archive ts_10319001v010301p0.zip which accompanies the present document. Each Huffman table consists of a sub-table containing length values and another sub-table containing the codeword. The indexing starts with index 0 and ends with index codebook_length - 1.
A.1 ASF Huffman codebook tables Table A.1: ASF scale factor Huffman codebook
Codebook name ASF_HCB_SCALEFAC Codebook length table ASF_HCB_SCALEFAC_LEN Codebook codeword table ASF_HCB_SCALEFAC_CW codebook_length 121
Table A.2: ASF spectrum Huffman codebook 1
Codebook name ASF_HCB_1 Codebook length table ASF_HCB_1_LEN Codebook codeword table ASF_HCB_1_CW codebook_length 81 cb_mod 3 cb_mod2 9 cb_mod3 27 cb_off 1
Table A.3: ASF spectrum Huffman codebook 2
Codebook name ASF_HCB_2 Codebook length table ASF_HCB_2_LEN Codebook codeword table ASF_HCB_2_CW codebook_length 81 cb_mod 3 cb_mod2 9 cb_mod3 27 cb_off 1
Table A.4: ASF spectrum Huffman codebook 3
Codebook name ASF_HCB_3 Codebook length table ASF_HCB_3_LEN Codebook codeword table ASF_HCB_3_CW codebook_length 81 cb_mod 3 cb_mod2 9 cb_mod3 27 cb_off 0
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)249
Table A.5: ASF spectrum Huffman codebook 4
Codebook name ASF_HCB_4 Codebook length table ASF_HCB_4_LEN Codebook codeword table ASF_HCB_4_CW codebook_length 81 cb_mod 3 cb_mod2 9 cb_mod3 27 cb_off 0
Table A.6: ASF spectrum Huffman codebook 5
Codebook name ASF_HCB_5 Codebook length table ASF_HCB_5_LEN Codebook codeword table ASF_HCB_5_CW codebook_length 81 cb_mod 9 cb_off 4
Table A.7: ASF spectrum Huffman codebook 6
Codebook name ASF_HCB_6 Codebook length table ASF_HCB_6_LEN Codebook codeword table ASF_HCB_6_CW codebook_length 81 cb_mod 9 cb_off 4
Table A.8: ASF spectrum Huffman codebook 7
Codebook name ASF_HCB_7 Codebook length table ASF_HCB_7_LEN Codebook codeword table ASF_HCB_7_CW codebook_length 64 cb_mod 8 cb_off 0
Table A.9: ASF spectrum Huffman codebook 8
Codebook name ASF_HCB_8 Codebook length table ASF_HCB_8_LEN Codebook codeword table ASF_HCB_8_CW codebook_length 64 cb_mod 8 cb_off 0
Table A.10: ASF spectrum Huffman codebook 9
Codebook name ASF_HCB_9 Codebook length table ASF_HCB_9_LEN Codebook codeword table ASF_HCB_9_CW codebook_length 169 cb_mod 13 cb_off 0
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)250
Table A.11: ASF spectrum Huffman codebook 10
Codebook name ASF_HCB_10 Codebook length table ASF_HCB_10_LEN Codebook codeword table ASF_HCB_10_CW codebook_length 169 cb_mod 13 cb_off 0
Table A.12: ASF spectrum Huffman codebook 11
Codebook name ASF_HCB_11 Codebook length table ASF_HCB_11_LEN Codebook codeword table ASF_HCB_11_CW codebook_length 289 cb_mod 17 cb_off 0
Table A.13: A-SPX SNF Huffman codebook
Codebook name ASF_HCB_SNF Codebook length table ASF_HCB_SNF_LEN Codebook codeword table ASF_HCB_SNF_CW codebook_length 22
Table A.14: CB_DIM
Codebook number 1 2 3 4 5 6 7 8 9 10 11 CB_DIM 4 4 4 4 2 2 2 2 2 2 2
Table A.15: UNSIGNED_CB
Codebook number 1 2 3 4 5 6 7 8 9 10 11 UNSIGNED_CB false false true true false false true true true true true
A.2 A-SPX Huffman codebook tables Table A.16: A-SPX Huffman codebook ASPX_HCB_ENV_LEVEL_15_F0
Codebook name ASPX_HCB_ENV_LEVEL_15_F0 Codebook length table ASPX_HCB_ENV_LEVEL_15_F0_LEN Codebook codeword table ASPX_HCB_ENV_LEVEL_15_F0_CW codebook_length 71
Table A.17: A-SPX Huffman codebook ASPX_HCB_ENV_LEVEL_15_DF
Codebook name ASPX_HCB_ENV_LEVEL_15_DF Codebook length table ASPX_HCB_ENV_LEVEL_15_DF_LEN Codebook codeword table ASPX_HCB_ENV_LEVEL_15_DF_CW codebook_length 141 cb_off 70
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)251
Table A.18: A-SPX Huffman codebook ASPX_HCB_ENV_LEVEL_15_DT
Codebook name ASPX_HCB_ENV_LEVEL_15_DT Codebook length table ASPX_HCB_ENV_LEVEL_15_DT_LEN Codebook codeword table ASPX_HCB_ENV_LEVEL_15_DT_CW codebook_length 141 cb_off 70
Table A.19: A-SPX Huffman codebook ASPX_HCB_ENV_BALANCE_15_F0
Codebook name ASPX_HCB_ENV_BALANCE_15_F0 Codebook length table ASPX_HCB_ENV_BALANCE_15_F0_LEN Codebook codeword table ASPX_HCB_ENV_BALANCE_15_F0_CW codebook_length 25
Table A.20: A-SPX Huffman codebook ASPX_HCB_ENV_BALANCE_15_DF
Codebook name ASPX_HCB_ENV_BALANCE_15_DF Codebook length table ASPX_HCB_ENV_BALANCE_15_DF_LEN Codebook codeword table ASPX_HCB_ENV_BALANCE_15_DF_CW codebook_length 49 cb_off 24
Table A.21: A-SPX Huffman codebook ASPX_HCB_ENV_BALANCE_15_DT
Codebook name ASPX_HCB_ENV_BALANCE_15_DT Codebook length table ASPX_HCB_ENV_BALANCE_15_DT_LEN Codebook codeword table ASPX_HCB_ENV_BALANCE_15_DT_CW codebook_length 49 cb_off 24
Table A.22: A-SPX Huffman codebook ASPX_HCB_ENV_LEVEL_30_F0
Codebook name ASPX_HCB_ENV_LEVEL_30_F0 Codebook length table ASPX_HCB_ENV_LEVEL_30_F0_LEN Codebook codeword table ASPX_HCB_ENV_LEVEL_30_F0_CW codebook_length 36
Table A.23: A-SPX Huffman codebook ASPX_HCB_ENV_LEVEL_30_DF
Codebook name ASPX_HCB_ENV_LEVEL_30_DF Codebook length table ASPX_HCB_ENV_LEVEL_30_DF_LEN Codebook codeword table ASPX_HCB_ENV_LEVEL_30_DF_CW codebook_length 71 cb_off 35
Table A.24: A-SPX Huffman codebook ASPX_HCB_ENV_LEVEL_30_DT
Codebook name ASPX_HCB_ENV_LEVEL_30_DT Codebook length table ASPX_HCB_ENV_LEVEL_30_DT_LEN Codebook codeword table ASPX_HCB_ENV_LEVEL_30_DT_CW codebook_length 71 cb_off 35
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)252
Table A.25: A-SPX Huffman codebook ASPX_HCB_ENV_BALANCE_30_F0
Codebook name ASPX_HCB_ENV_BALANCE_30_F0 Codebook length table ASPX_HCB_ENV_BALANCE_30_F0_LEN Codebook codeword table ASPX_HCB_ENV_BALANCE_30_F0_CW codebook_length 13
Table A.26: A-SPX Huffman codebook ASPX_HCB_ENV_BALANCE_30_DF
Codebook name ASPX_HCB_ENV_BALANCE_30_DF Codebook length table ASPX_HCB_ENV_BALANCE_30_DF_LEN Codebook codeword table ASPX_HCB_ENV_BALANCE_30_DF_CW codebook_length 25 cb_off 12
Table A.27: A-SPX Huffman codebook ASPX_HCB_ENV_BALANCE_30_DT
Codebook name ASPX_HCB_ENV_BALANCE_30_DT Codebook length table ASPX_HCB_ENV_BALANCE_30_DT_LEN Codebook codeword table ASPX_HCB_ENV_BALANCE_30_DT_CW codebook_length 25 cb_off 12
Table A.28: A-SPX Huffman codebook ASPX_HCB_NOISE_LEVEL_F0
Codebook name ASPX_HCB_NOISE_LEVEL_F0 Codebook length table ASPX_HCB_NOISE_LEVEL_F0_LEN Codebook codeword table ASPX_HCB_NOISE_LEVEL_F0_CW codebook_length 30
Table A.29: A-SPX Huffman codebook ASPX_HCB_NOISE_LEVEL_DF
Codebook name ASPX_HCB_NOISE_LEVEL_DF Codebook length table ASPX_HCB_NOISE_LEVEL_DF_LEN Codebook codeword table ASPX_HCB_NOISE_LEVEL_DF_CW codebook_length 59 cb_off 29
Table A.30: A-SPX Huffman codebook ASPX_HCB_NOISE_LEVEL_DT
Codebook name ASPX_HCB_NOISE_LEVEL_DT Codebook length table ASPX_HCB_NOISE_LEVEL_DT_LEN Codebook codeword table ASPX_HCB_NOISE_LEVEL_DT_CW codebook_length 59 cb_off 29
Table A.31: A-SPX Huffman codebook ASPX_HCB_NOISE_BALANCE_F0
Codebook name ASPX_HCB_NOISE_BALANCE_F0 Codebook length table ASPX_HCB_NOISE_BALANCE_F0_LEN Codebook codeword table ASPX_HCB_NOISE_BALANCE_F0_CW codebook_length 13
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)253
Table A.32: A-SPX Huffman codebook ASPX_HCB_NOISE_BALANCE_DF
Codebook name ASPX_HCB_NOISE_BALANCE_DF Codebook length table ASPX_HCB_NOISE_BALANCE_DF_LEN Codebook codeword table ASPX_HCB_NOISE_BALANCE_DF_CW codebook_length 25 cb_off 12
Table A.33: A-SPX Huffman codebook ASPX_HCB_NOISE_BALANCE_DT
Codebook name ASPX_HCB_NOISE_BALANCE_DT Codebook length table ASPX_HCB_NOISE_BALANCE_DT_LEN Codebook codeword table ASPX_HCB_NOISE_BALANCE_DT_CW codebook_length 25 cb_off 12
A.3 A-CPL Huffman codebook tables Table A.34: A-CPL Huffman codebook ACPL_HCB_ALPHA_COARSE_F0
Codebook name ACPL_HCB_ALPHA_COARSE_F0 Codebook length table ACPL_HCB_ALPHA_COARSE_F0_LEN Codebook codeword table ACPL_HCB_ALPHA_COARSE_F0_CW codebook_length 17 cb_off 0
Table A.35: A-CPL Huffman codebook ACPL_HCB_ALPHA_FINE_F0
Codebook name ACPL_HCB_ALPHA_FINE_F0 Codebook length table ACPL_HCB_ALPHA_FINE_F0_LEN Codebook codeword table ACPL_HCB_ALPHA_FINE_F0_CW codebook_length 33 cb_off 0
Table A.36: A-CPL Huffman codebook ACPL_HCB_ALPHA_COARSE_DF
Codebook name ACPL_HCB_ALPHA_COARSE_DF Codebook length table ACPL_HCB_ALPHA_COARSE_DF_LEN Codebook codeword table ACPL_HCB_ALPHA_COARSE_DF_CW codebook_length 33 cb_off 16
Table A.37: A-CPL Huffman codebook ACPL_HCB_ALPHA_FINE_DF
Codebook name ACPL_HCB_ALPHA_FINE_DF Codebook length table ACPL_HCB_ALPHA_FINE_DF_LEN Codebook codeword table ACPL_HCB_ALPHA_FINE_DF_CW codebook_length 65 cb_off 32
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)254
Table A.38: A-CPL Huffman codebook ACPL_HCB_ALPHA_COARSE_DT
Codebook name ACPL_HCB_ALPHA_COARSE_DT Codebook length table ACPL_HCB_ALPHA_COARSE_DT_LEN Codebook codeword table ACPL_HCB_ALPHA_COARSE_DT_CW codebook_length 33 cb_off 16
Table A.39: A-CPL Huffman codebook ACPL_HCB_ALPHA_FINE_DT
Codebook name ACPL_HCB_ALPHA_FINE_DT Codebook length table ACPL_HCB_ALPHA_FINE_DT_LEN Codebook codeword table ACPL_HCB_ALPHA_FINE_DT_CW codebook_length 65 cb_off 32
Table A.40: A-CPL Huffman codebook ACPL_HCB_BETA_COARSE_F0
Codebook name ACPL_HCB_BETA_COARSE_F0 Codebook length table ACPL_HCB_BETA_COARSE_F0_LEN Codebook codeword table ACPL_HCB_BETA_COARSE_F0_CW codebook_length 5 cb_off 0
Table A.41: A-CPL Huffman codebook ACPL_HCB_BETA_FINE_F0
Codebook name ACPL_HCB_BETA_FINE_F0 Codebook length table ACPL_HCB_BETA_FINE_F0_LEN Codebook codeword table ACPL_HCB_BETA_FINE_F0_CW codebook_length 9 cb_off 0
Table A.42: A-CPL Huffman codebook ACPL_HCB_BETA_COARSE_DF
Codebook name ACPL_HCB_BETA_COARSE_DF Codebook length table ACPL_HCB_BETA_COARSE_DF_LEN Codebook codeword table ACPL_HCB_BETA_COARSE_DF_CW codebook_length 9 cb_off 4
Table A.43: A-CPL Huffman codebook ACPL_HCB_BETA_FINE_DF
Codebook name ACPL_HCB_BETA_FINE_DF Codebook length table ACPL_HCB_BETA_FINE_DF_LEN Codebook codeword table ACPL_HCB_BETA_FINE_DF_CW codebook_length 17 cb_off 8
Table A.44: A-CPL Huffman codebook ACPL_HCB_BETA_COARSE_DT
Codebook name ACPL_HCB_BETA_COARSE_DT Codebook length table ACPL_HCB_BETA_COARSE_DT_LEN Codebook codeword table ACPL_HCB_BETA_COARSE_DT_CW codebook_length 9 cb_off 4
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)255
Table A.45: A-CPL Huffman codebook ACPL_HCB_BETA_FINE_DT
Codebook name ACPL_HCB_BETA_FINE_DT Codebook length table ACPL_HCB_BETA_FINE_DT_LEN Codebook codeword table ACPL_HCB_BETA_FINE_DT_CW codebook_length 17 cb_off 8
Table A.46: A-CPL Huffman codebook ACPL_HCB_BETA3_COARSE_F0
Codebook name ACPL_HCB_BETA3_COARSE_F0 Codebook length table ACPL_HCB_BETA3_COARSE_F0_LEN Codebook codeword table ACPL_HCB_BETA3_COARSE_F0_CW codebook_length 9 cb_off 0
Table A.47: A-CPL Huffman codebook ACPL_HCB_BETA3_FINE_F0
Codebook name ACPL_HCB_BETA3_FINE_F0 Codebook length table ACPL_HCB_BETA3_FINE_F0_LEN Codebook codeword table ACPL_HCB_BETA3_FINE_F0_CW codebook_length 17 cb_off 0
Table A.48: A-CPL Huffman codebook ACPL_HCB_BETA3_COARSE_DF
Codebook name ACPL_HCB_BETA3_COARSE_DF Codebook length table ACPL_HCB_BETA3_COARSE_DF_LEN Codebook codeword table ACPL_HCB_BETA3_COARSE_DF_CW codebook_length 17 cb_off 8
Table A.49: A-CPL Huffman codebook ACPL_HCB_BETA3_FINE_DF
Codebook name ACPL_HCB_BETA3_FINE_DF Codebook length table ACPL_HCB_BETA3_FINE_DF_LEN Codebook codeword table ACPL_HCB_BETA3_FINE_DF_CW codebook_length 33 cb_off 16
Table A.50: A-CPL Huffman codebook ACPL_HCB_BETA3_COARSE_DT
Codebook name ACPL_HCB_BETA3_COARSE_DT Codebook length table ACPL_HCB_BETA3_COARSE_DT_LEN Codebook codeword table ACPL_HCB_BETA3_COARSE_DT_CW codebook_length 17 cb_off 8
Table A.51: A-CPL Huffman codebook ACPL_HCB_BETA3_FINE_DT
Codebook name ACPL_HCB_BETA3_FINE_DT Codebook length table ACPL_HCB_BETA3_FINE_DT_LEN Codebook codeword table ACPL_HCB_BETA3_FINE_DT_CW codebook_length 33 cb_off 16
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)256
Table A.52: A-CPL Huffman codebook ACPL_HCB_GAMMA_COARSE_F0
Codebook name ACPL_HCB_GAMMA_COARSE_F0 Codebook length table ACPL_HCB_GAMMA_COARSE_F0_LEN Codebook codeword table ACPL_HCB_GAMMA_COARSE_F0_CW codebook_length 21 cb_off 10
Table A.53: A-CPL Huffman codebook ACPL_HCB_GAMMA_FINE_F0
Codebook name ACPL_HCB_GAMMA_FINE_F0 Codebook length table ACPL_HCB_GAMMA_FINE_F0_LEN Codebook codeword table ACPL_HCB_GAMMA_FINE_F0_CW codebook_length 41 cb_off 20
Table A.54: A-CPL Huffman codebook ACPL_HCB_GAMMA_COARSE_DF
Codebook name ACPL_HCB_GAMMA_COARSE_DF Codebook length table ACPL_HCB_GAMMA_COARSE_DF_LEN Codebook codeword table ACPL_HCB_GAMMA_COARSE_DF_CW codebook_length 41 cb_off 20
Table A.55: A-CPL Huffman codebook ACPL_HCB_GAMMA_FINE_DF
Codebook name ACPL_HCB_GAMMA_FINE_DF Codebook length table ACPL_HCB_GAMMA_FINE_DF_LEN Codebook codeword table ACPL_HCB_GAMMA_FINE_DF_CW codebook_length 81 cb_off 40
Table A.56: A-CPL Huffman codebook ACPL_HCB_GAMMA_COARSE_DT
Codebook name ACPL_HCB_GAMMA_COARSE_DT Codebook length table ACPL_HCB_GAMMA_COARSE_DT_LEN Codebook codeword table ACPL_HCB_GAMMA_COARSE_DT_CW codebook_length 41 cb_off 20
Table A.57: A-CPL Huffman codebook ACPL_HCB_GAMMA_FINE_DT
Codebook name ACPL_HCB_GAMMA_FINE_DT Codebook length table ACPL_HCB_GAMMA_FINE_DT_LEN Codebook codeword table ACPL_HCB_GAMMA_FINE_DT_CW codebook_length 81 cb_off 40
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)257
A.4 Dialogue enhancement Huffman codebook tables Table A.58: Dialogue enhancement Huffman codebook DE_HCB_ABS_0
Codebook name DE_HCB_ABS_0 Codebook length table DE_HCB_ABS_0_LEN Codebook codeword table DE_HCB_ABS_0_CW codebook_length 32 cb_off 0
Table A.59: Dialogue enhancement Huffman codebook DE_HCB_DIFF_0
Codebook name DE_HCB_DIFF_0 Codebook length table DE_HCB_DIFF_0_LEN Codebook codeword table DE_HCB_DIFF_0_CW codebook_length 63 cb_off 31
Table A.60: Dialogue enhancement Huffman codebook DE_HCB_ABS_1
Codebook name DE_HCB_ABS_1 Codebook length table DE_HCB_ABS_1_LEN Codebook codeword table DE_HCB_ABS_1_CW codebook_length 61 cb_off 30
Table A.61: Dialogue enhancement Huffman codebook DE_HCB_DIFF_1
Codebook name DE_HCB_DIFF_1 Codebook length table DE_HCB_DIFF_1_LEN Codebook codeword table DE_HCB_DIFF_1_CW codebook_length 121 cb_off 60
A.5 Dynamic range control Huffman codebook table Table A.62: Dynamic range control Huffman codebook DRC_HCB
Codebook name DRC_HCB Codebook length table DRC_HCB_LEN Codebook codeword table DRC_HCB_CW codebook_length 255 cb_off 127
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)258
Annex B (normative): ASF scale factor band tables
Table B.1: Number of scale factor bands for 44,1 kHz or 48 kHz sampling frequency
transform length num_sfb 2 048 63 1 920 61 1 536 55 1 024 49 960 49 768 43 512 36 480 36 384 33 256 20 240 20 192 18 128 14 120 14 96 12
Table B.2: Number of scale factor bands for 96 kHz sampling frequency
transform length num_sfb 4 096 79 3 840 76 3 072 67 2 048 57 1 920 57 1 536 49 1 024 44 920 44 768 39 512 28 480 28 384 24 256 22 240 22 192 18
Table B.3: Number of scale factor bands for 192 kHz sampling frequency
transform length num_sfb 8 192 111 7 680 106 6 144 91 4 096 73 3 840 72 3 072 61 2 048 60 1 920 59 1 536 51 1 024 36 960 36 768 30 512 30 480 30 384 24
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)259
Table B.4: Scale factor band offsets for sampling frequency 44,1 kHz or 48 kHz and transform length 2 048, 1 920 or 1 536; or for sampling frequency 96 kHz and transform length 4 096, 3 840, or 3 072; or
for sampling frequency 192 kHz and transform length 8 192, 7 680, or 6 144
sfb sfb_offset - sfb sfb_offset 2 048@44,1 2 048@48 4 096@96
8 192@192
1 920@48 3 840@96
7 680@192
1 536@48 3 072@96
6 144@192
- 2 048@44,1 2 048@48 4 096@96
8 192@192
1 920@48 3 840@96
7 680@192
1 536@48 3 072@96
6 144@192
0 0 0 0 - 56 1 600 1 600 1 664 1 4 4 4 - 57 1 664 1 664 1 792 2 8 8 8 - 58 1 728 1 728 1 920 3 12 12 12 - 59 1 792 1 792 2 048 4 16 16 16 - 60 1 856 1 856 2 176 5 20 20 20 - 61 1 920 1 920 2 304 6 24 24 24 - 62 1 984 2 048 2 432 7 28 28 28 - 63 2 048 2 176 2 560 8 32 32 32 - 64 2 176 2 304 2 688 9 36 36 36 - 65 2 304 2 432 2 816 10 40 40 40 - 66 2 432 2 560 2 944 11 44 44 44 - 67 2 560 2 688 3 072 12 52 52 52 - 68 2 688 2 816 3 200 13 60 60 60 - 69 2 816 2 944 3 328 14 68 68 68 - 70 2 944 3 072 3 456 15 76 76 76 - 71 3 072 3 200 3 584 16 84 84 84 - 72 3 200 3 328 3 712 17 92 92 92 - 73 3 328 3 456 3 840 18 100 100 100 - 74 3 456 3 584 3 968 19 108 108 108 - 75 3 584 3 712 4 096 20 116 116 116 - 76 3 712 3 840 4 224 21 124 124 124 - 77 3 840 3 968 4 352 22 136 136 136 - 78 3 968 4 096 4 480 23 148 148 148 - 79 4 096 4 224 4 608 24 160 160 160 - 80 4 224 4 352 4 736 25 172 172 172 - 81 4 352 4 480 4 864 26 188 188 188 - 82 4 480 4 608 4 992 27 204 204 204 - 83 4 608 4 736 5 120 28 220 220 220 - 84 4 736 4 864 5 248 29 240 240 240 - 85 4 864 4 992 5 376 30 260 260 260 - 86 4 992 5 120 5 504 31 284 284 284 - 87 5 120 5 248 5 632 32 308 308 308 - 88 5 248 5 376 5 760 33 336 336 336 - 89 5 376 5 504 5 888 34 364 364 364 - 90 5 504 5 632 6 016 35 396 396 396 - 91 5 632 5 760 6 144 36 432 432 432 - 92 5 760 5 888 - 37 468 468 468 - 93 5 888 6 016 - 38 508 508 508 - 94 6 016 6 144 - 39 552 552 552 - 95 6 144 6 272 - 40 600 600 600 - 96 6 272 6 400 - 41 652 652 652 - 97 6 400 6 528 - 42 704 704 704 - 98 6 528 6 656 - 43 768 768 768 - 99 6 656 6 784 - 44 832 832 832 - 100 6 784 6 912 - 45 896 896 896 - 101 6 912 7 040 - 46 960 960 960 - 102 7 040 7 168 - 47 1 024 1 024 1 024 - 103 7 168 7 296 - 48 1 088 1 088 1 088 - 104 7 296 7 424 - 49 1 152 1 152 1 152 - 105 7 424 7 552 - 50 1 216 1 216 1 216 - 106 7 552 7 680 - 51 1 280 1 280 1 280 - 107 7 680 - - 52 1 344 1 344 1 344 - 108 7 808 - - 53 1 408 1 408 1 408 - 109 7 936 - - 54 1 472 1 472 1 472 - 110 8 064 - -
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)260
sfb sfb_offset - sfb sfb_offset 2 048@44,1 2 048@48 4 096@96
8 192@192
1 920@48 3 840@96
7 680@192
1 536@48 3 072@96
6 144@192
- 2 048@44,1 2 048@48 4 096@96
8 192@192
1 920@48 3 840@96
7 680@192
1 536@48 3 072@96
6 144@192
55 1 536 1 536 1 536 - 111 8 192 - -
Table B.5: Scale factor band offsets for sampling frequency 44,1 kHz or 48 kHz and transform length 1 024, 960, or 768; or for sampling frequency 96 kHz and transform length 2 048, 1 920, or 1 536; or
for sampling frequency 192 kHz and transform length 4 096, 3 840, or 3 072
sfb sfb_offset - sfb sfb_offset 1 024@44,1 1 024@48 2 048@96
4 096@192
960@48 1 920@96
3 840@192
768@48 1 536@96
3 072@192
- 1 024@44,1 1 024@48 2 048@96
4 096@192
960@48 1 920@96
3 840@192
768@48 1 536@96
3 072@192
0 0 0 0 - 37 576 576 576 1 4 4 4 - 38 608 608 608 2 8 8 8 - 39 640 640 640 3 12 12 12 - 40 672 672 672 4 16 16 16 - 41 704 704 704 5 20 20 20 - 42 736 736 736 6 24 24 24 - 43 768 768 768 7 28 28 28 - 44 800 800 896 8 32 32 32 - 45 832 832 1 024 9 36 36 36 - 46 864 864 1 152 10 40 40 40 - 47 896 896 1 280 11 48 48 48 - 48 928 928 1 408 12 56 56 56 - 49 1 024 960 1 536 13 64 64 64 - 50 1 152 1 024 1 664 14 72 72 72 - 51 1 280 1 152 1 792 15 80 80 80 - 52 1 408 1 280 1 920 16 88 88 88 - 53 1 536 1 408 2 048 17 96 96 96 - 54 1 664 1 536 2 176 18 108 108 108 - 55 1 792 1 664 2 304 19 120 120 120 - 56 1 920 1 792 2 432 20 132 132 132 - 57 2 048 1 920 2 560 21 144 144 144 - 58 2 176 2 048 2 688 22 160 160 160 - 59 2 304 2 176 2 816 23 176 176 176 - 60 2 432 2 304 2 944 24 196 196 196 - 61 2 560 2 432 3 072 25 216 216 216 - 62 2 688 2 560 - 26 240 240 240 - 63 2 816 2 688 - 27 264 264 264 - 64 2 944 2 816 - 28 292 292 292 - 65 3 072 2 944 - 29 320 320 320 - 66 3 200 3 072 - 30 352 352 352 - 67 3 328 3 200 - 31 384 384 384 - 68 3 456 3 328 - 32 416 416 416 - 69 3 584 3 456 - 33 448 448 448 - 70 3 712 3 584 - 34 480 480 480 - 71 3 840 3 712 - 35 512 512 512 - 72 3 968 3 840 - 36 544 544 544 - 73 4 096 - -
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)261
Table B.6: Scale factor band offsets for sampling frequency 44,1 kHz or 48 kHz and transform length 512, 480, or 384; or for sampling frequency 96 kHz and transform length 1 024, 960, or 768 or for
sampling frequency 192 kHz and transform length 2 048, 1 920, or 1 536
sfb sfb_offset - sfb sfb_offset 512@44,1 512@48
1 024@96 2 048@192
480@48 960@96
1 920@192
384@48 768@96
1 536@192
- 512@44,1 512@48
1 024@96 2 048@192
480@48 960@96
1 920@192
384@48 768@96
1 536@192
0 0 0 0 - 31 332 332 332 1 4 4 4 - 32 364 364 364 2 8 8 8 - 33 396 396 384 3 12 12 12 - 34 428 428 448 4 16 16 16 - 35 460 460 512 5 20 20 20 - 36 512 480 576 6 24 24 24 - 37 576 512 640 7 28 28 28 - 38 640 576 704 8 32 32 32 - 39 704 640 768 9 36 36 36 - 40 768 704 832 10 40 40 40 - 41 832 768 896 11 44 44 44 - 42 896 832 960 12 48 48 48 - 43 960 896 1 024 13 52 52 52 - 44 1 024 960 1 088 14 56 56 56 - 45 1 088 1 024 1 152 15 60 60 60 - 46 1 152 1 088 1 216 16 68 68 68 - 47 1 216 1 152 1 280 17 76 76 76 - 48 1 280 1 216 1 344 18 84 84 84 - 49 1 344 1 280 1 408 19 92 92 92 - 50 1 408 1 344 1 472 20 100 100 100 - 51 1 472 1 408 1 536 21 112 112 112 - 52 1 536 1 472 - 22 124 124 124 - 53 1 600 1 536 - 23 136 136 136 - 54 1 664 1 600 - 24 148 148 148 - 55 1 728 1 664 - 25 164 164 164 - 56 1 792 1 728 - 26 184 184 184 - 57 1 856 1 792 - 27 208 208 208 - 58 1 920 1 856 - 28 236 236 236 - 59 1 984 1 920 - 29 268 268 268 - 60 2 048 - - 30 300 300 300 - - - - -
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)262
Table B.7: Scale factor band offsets for sampling frequency 44,1 kHz or 48 kHz and transform length 256, 240, 192, 128, 120, or 96; or for sampling frequency 96 kHz and transform length 512, 480, 384,
256, 240, or 192; or for sampling frequency 192 kHz and transform length 1 024, 960, 768, 512, 480, or 384
sfb sfb_offset 256@44,1
256@48 512@96 1 024@192
240@48 480@96 960@192
192@48 384@96 768@192
128@44,1 128@48 256@96 512@192
120@48 240@96 480@192
96@48 192@96
384@192 0 0 0 0 0 0 0 1 4 4 4 4 4 4 2 8 8 8 8 8 8 3 12 12 12 12 12 12 4 16 16 16 16 16 16 5 20 20 20 20 20 20 6 24 24 24 28 28 28 7 28 28 28 36 36 36 8 36 36 36 44 44 44 9 44 44 44 56 56 56 10 52 52 52 68 68 68 11 64 64 64 80 80 80 12 76 76 76 96 96 96 13 92 92 92 112 112 112 14 108 108 108 128 120 128 15 128 128 128 144 128 144 16 148 148 148 160 144 160 17 172 172 172 176 160 176 18 196 196 192 192 176 192 19 224 224 224 208 192 224 20 256 240 256 224 208 256 21 288 256 288 240 224 288 22 320 288 320 256 240 320 23 352 320 352 288 256 352 24 384 352 384 320 288 384 25 416 384 448 352 320 - 26 448 416 512 384 352 - 27 480 448 576 416 384 - 28 512 480 640 448 416 - 29 576 512 704 480 448 - 30 640 576 768 512 480 - 31 704 640 - - - - 32 768 704 - - - - 33 832 768 - - - - 34 896 832 - - - - 35 960 896 - - - - 36 1 024 960 - - - -
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)263
Table B.8: Mapping from max_sfb_master from transform length 2 048 to different transform lengths
max_sfb_master[2 048] n_sfb_side[1 024] n_sfb_side[512] n_sfb_side[256] n_sfb_side[128] 0 0 0 0 0 1 1 1 1 1 2 1 1 1 1 3 2 1 1 1 4 2 1 1 1 5 3 2 1 1 6 3 2 1 1 7 4 2 1 1 8 4 2 1 1 9 5 3 2 1 10 5 3 2 1 11 6 3 2 1 12 7 4 2 1 13 8 4 2 1 14 9 5 3 2 15 10 5 3 2 16 11 6 3 2 17 11 6 3 2 18 12 7 4 2 19 12 7 4 2 20 13 8 4 2 21 13 8 4 2 22 14 9 5 3 23 15 10 5 3 24 15 10 5 3 25 16 11 6 3 26 17 12 6 3 27 18 13 7 4 28 19 14 7 4 29 19 15 8 4 30 20 16 8 5 31 21 17 8 5
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)264
Table B.9: Mapping from max_sfb_master from transform length 1 024 to different transform lengths
max_sfb_master[1 024] n_sfb_side[512] n_sfb_side[256] n_sfb_side[128] 0 0 0 0 1 1 1 1 2 1 1 1 3 2 1 1 4 2 1 1 5 3 2 1 6 3 2 1 7 4 2 1 8 4 2 1 9 5 3 2 10 5 3 2 11 6 3 2 12 7 4 2 13 8 4 2 14 9 5 3 15 10 5 3 16 11 6 3 17 12 6 3 18 14 7 4 19 15 8 4 20 16 8 5 21 17 8 5 22 18 9 5 23 19 9 6 24 20 10 6 25 21 11 6 26 22 11 7 27 23 12 7 28 24 12 8 29 25 13 8 30 26 13 8 31 27 14 9
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)265
Table B.10: Mapping from max_sfb_master from transform length 512
max_sfb_master[512] n_sfb_side[256] n_sfb_side[128] 0 0 0 1 1 1 2 1 1 3 2 1 4 2 1 5 3 2 6 3 2 7 4 2 8 4 2 9 5 3 10 5 3 11 6 3 12 6 3 13 7 4 14 7 4 15 8 4 16 8 5 17 9 5 18 9 6 19 10 6 20 10 6 21 11 6 22 11 7 23 12 7 24 12 8 25 13 8 26 13 9 27 14 9 28 15 10 29 16 10 30 17 11 31 17 12
Table B.11: Mapping from max_sfb_master from transform length 256
max_sfb_master[256] n_sfb_side[128] 0 0 1 1 2 1 3 2 4 2 5 3 6 3 7 4 8 5 9 6 10 6 11 7 12 8 13 9 14 9 15 10
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)266
Table B.12: Mapping from max_sfb_master from transform length 1 920 to different transform lengths
max_sfb_master[1 920] n_sfb_side[960] n_sfb_side[480] n_sfb_side[240] n_sfb_side[120] 0 0 0 0 0 1 1 1 1 1 2 1 1 1 1 3 2 1 1 1 4 2 1 1 1 5 3 2 1 1 6 3 2 1 1 7 4 2 1 1 8 4 2 1 1 9 5 3 2 1 10 5 3 2 1 11 6 3 2 1 12 7 4 2 1 13 8 4 2 1 14 9 5 3 2 15 10 5 3 2 16 11 6 3 2 17 11 6 3 2 18 12 7 4 2 19 12 7 4 2 20 13 8 4 2 21 13 8 4 2 22 14 9 5 3 23 15 10 5 3 24 15 10 5 3 25 16 11 6 3 26 17 12 6 3 27 18 13 7 4 28 19 14 7 4 29 19 15 8 4 30 20 16 8 5 31 21 17 8 5
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)267
Table B.13: Mapping from max_sfb_master from transform length 960 to different transform lengths
max_sfb_master[960] n_sfb_side[480] n_sfb_side[240] n_sfb_side[120] 0 0 0 0 1 1 1 1 2 1 1 1 3 2 1 1 4 2 1 1 5 3 2 1 6 3 2 1 7 4 2 1 8 4 2 1 9 5 3 2 10 5 3 2 11 6 3 2 12 7 4 2 13 8 4 2 14 9 5 3 15 10 5 3 16 11 6 3 17 12 6 3 18 14 7 4 19 15 8 4 20 16 8 5 21 17 8 5 22 18 9 5 23 19 9 6 24 20 10 6 25 21 11 6 26 22 11 7 27 23 12 7 28 24 12 8 29 25 13 8 30 26 13 8 31 27 14 9
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)268
Table B.14: Mapping from max_sfb_master from transform length 480
max_sfb_master[480] n_sfb_side[240] n_sfb_side[120] 0 0 0 1 1 1 2 1 1 3 2 1 4 2 1 5 3 2 6 3 2 7 4 2 8 4 2 9 5 3 10 5 3 11 6 3 12 6 3 13 7 4 14 7 4 15 8 4 16 8 5 17 9 5 18 9 6 19 10 6 20 10 6 21 11 6 22 11 7 23 12 7 24 12 8 25 13 8 26 13 9 27 14 9 28 15 10 29 16 10 30 17 11 31 17 12
Table B.15: Mapping from max_sfb_master from transform length 240
max_sfb_master[240] n_sfb_side[120] 0 0 1 1 2 1 3 2 4 2 5 3 6 3 7 4 8 5 9 6 10 6 11 7 12 8 13 9 14 9 15 10
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)269
Table B.16: Mapping from max_sfb_master from transform length 1 536 to different transform lengths
max_sfb_master[1 536] n_sfb_side[768] n_sfb_side[384] n_sfb_side[192] n_sfb_side[96] 0 0 0 0 0 1 1 1 1 1 2 1 1 1 1 3 2 1 1 1 4 2 1 1 1 5 3 2 1 1 6 3 2 1 1 7 4 2 1 1 8 4 2 1 1 9 5 3 2 1 10 5 3 2 1 11 6 3 2 1 12 7 4 2 1 13 8 4 2 1 14 9 5 3 2 15 10 5 3 2 16 11 6 3 2 17 11 6 3 2 18 12 7 4 2 19 12 7 4 2 20 13 8 4 2 21 13 8 4 2 22 14 9 5 3 23 15 10 5 3 24 15 10 5 3 25 16 11 6 3 26 17 12 6 3 27 18 13 7 4 28 19 14 7 4 29 19 15 8 4 30 20 16 8 5 31 21 17 8 5
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)270
Table B.17: Mapping from max_sfb_master from transform length 768 to different transform lengths
max_sfb_master[768] n_sfb_side[384] n_sfb_side[192] n_sfb_side[96] 0 0 0 0 1 1 1 1 2 1 1 1 3 2 1 1 4 2 1 1 5 3 2 1 6 3 2 1 7 4 2 1 8 4 2 1 9 5 3 2 10 5 3 2 11 6 3 2 12 7 4 2 13 8 4 2 14 9 5 3 15 10 5 3 16 11 6 3 17 12 6 3 18 14 7 4 19 15 8 4 20 16 8 5 21 17 8 5 22 18 9 5 23 19 9 6 24 20 10 6 25 21 11 6 26 22 11 7 27 23 12 7 28 24 12 8 29 25 13 8 30 26 13 8 31 27 14 9
Table B.18: Mapping from max_sfb_master from transform length 384
max_sfb_master[384] n_sfb_side[192] n_sfb_side[96] 0 0 0 1 1 1 2 1 1 3 2 1 4 2 1 5 3 2 6 3 2 7 4 2 8 4 2 9 5 3 10 5 3 11 6 3 12 6 3 13 7 4 14 7 4 15 8 4
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)271
Table B.19: Mapping from max_sfb_master from transform length 192
max_sfb_master[192] n_sfb_side[96] 0 0 1 1 2 1 3 2 4 2 5 3 6 3 7 4
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)272
Annex C (normative): Speech spectral frontend tables
C.1 SSF bandwidths Table C.1 specifies the SSF bandwidths (number of spectral lines per band) for the different SSF block lengths.
Table C.1: SSF bandwidths
Bandindex Number of bins Block length 192
Block length 240
Block length 256
Block length 384
Block length 512
Block length 768
Block length 960
Block length 1 024
0 2 3 3 5 6 9 11 12 1 2 3 3 5 6 9 11 12 2 2 3 3 5 6 9 11 12 3 2 3 3 5 6 9 11 12 4 2 3 3 5 6 9 11 12 5 2 3 3 5 6 9 11 12 6 2 3 3 5 6 9 11 12 7 2 3 3 5 6 9 11 12 8 2 3 3 5 6 9 11 12 9 2 3 3 5 6 9 11 12 10 3 4 4 6 8 12 15 16 11 3 4 4 6 8 12 15 16 12 3 4 4 6 8 12 15 16 13 4 5 5 8 10 15 19 20 14 4 5 5 8 10 15 19 20 15 5 6 6 9 12 18 23 24 16 5 7 7 11 14 21 26 28 17 5 7 7 11 14 21 26 28 18 6 8 8 12 16 24 30 32
C.2 POST_GAIN_LUT Table C.2: POST_GAIN_LUT
Table name POST_GAIN_LUT table_length 20
C.3 PRED_GAIN_QUANT_TAB Table C.3: PRED_GAIN_QUANT_TAB[ ]
Table name PRED_GAIN_QUANT_TAB table_length 32
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)273
C.4 PRED_RFS_TABLE Table C.4: PRED_RFS_TABLE[ ]
Table name PRED_RFS_TABLE table_length 37
C.5 PRED_RTS_TABLE Table C.5: PRED_RTS_TABLE[ ]
Table name PRED_RTS_TABLE table_length 37
C.6 Quantized prediction coefficients The quantized prediction coefficients are stored in tables available available in the file ts_103190_tables.c contained in archive ts_10319001v010301p0.zip which accompanies the present document and are named ssf_pred_coeff_mat0[ ], …, ssf_pred_coeff_mat36[ ]. The mapping to the table PRED_COEFF_QUANT_MAT[tab_idx][ν][η][k], which is used in clause 5.2.8.1, is given by the following pseudocode.
Pseudocode 123
// Mapping of ssf_pred_coeff_mat<tab_idx> to PRED_COEFF_QUANT_MAT[tab_idx][ν][η][k] rfs = PRED_RFS_TABLE[tab_idx]; rts = PRED_RTS_TABLE[tab_idx]; table_index = (ν + rfs)*rts*33 + k*33 + η; PRED_COEFF_QUANT_MAT[tab_idx][ν][η][k] = ssf_pred_coeff_mat<tab_idx>[table_index];
C.7 CDF_TABLE Table C.6: CDF_TABLE
Table name CDF_TABLE table_length 705
C.8 PREDICTOR_GAIN_CDF_LUT Table C.7: PREDICTOR_GAIN_CDF_LUT
Table name PREDICTOR_GAIN_CDF_LUT table_length 33
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)274
C.9 ENVELOPE_CDF_LUT Table C.8: ENVELOPE_CDF_LUT
Table name ENVELOPE_CDF_LUT table_length 33
C.10 DITHER_TABLE Table C.9: DITHER_TABLE
Table name DITHER_TABLE table_length 256
C.11 RANDOM_NOISE_TABLE Table C.10: RANDOM_NOISE_TABLE
Table name RANDOM_NOISE_TABLE table_length 256
C.12 STEP_SIZES_Q4_15 Table C.11: STEP_SIZES_Q4_15
Table name STEP_SIZES_Q4_15 table_length 21
C.13 AC_COEFF_MAX_INDEX Table C.12: AC_COEFF_MAX_INDEX
Table name AC_COEFF_MAX_INDEX table_length 21
C.14 dB conversion tables Table C.13: SLOPES_DB_TO_LIN
Table name SLOPES_DB_TO_LIN table_length 10
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)275
Table C.14: OFFSETS_DB_TO_LIN
Table name OFFSETS_DB_TO_LIN table_length 10
Table C.15: SLOPES_LIN_TO_DB
Table name SLOPES_LIN_TO_DB table_length 50
Table C.16: OFFSETS_LIN_TO_DB
Table name OFFSETS_LIN_TO_DB table_length 50
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)276
Annex D (normative): Other tables
D.1 Channel names Table D.1 explains channel abbreviations used in the present document.
Table D.1: Channel abbreviations
Channel name Abbreviation Left L Centre C Right R Left Surround Ls Right Surround Rs Left Centre Lc Right Centre Rc Left Rear Surround Lrs Right Rear Surround Rrs Centre Surround Cs Top Surround Ts Left Surround Direct Lsd Right Surround Direct Rsd Left Wide Lw Right Wide Rw Left Vertical Height Vhl Right Vertical Height Vhr Centre Vertical Height Vhc Low-Frequency Effects LFE Low-Frequency Effects 2 LFE2
D.2 A-SPX noise table The A-SPX noise table referenced in table D.2 is available in the file ts_103190_tables.c contained in archive ts_10319001v010301p0.zip which accompanies the present document.
Table D.2: A-SPX noise table
Table name ASPX_NOISE num_columns 2 num_rows 512
D.3 QMF coefficients The Quadrature Mirror Filter (QMF) window coefficients referenced in table D.3 are available in the file ts_103190_tables.c contained in archive ts_10319001v010301p0.zip which accompanies the present document.
Table D.3: QWIN
Table name QWIN table_length 640
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)277
Annex E (normative): AC-4 bitstream storage in the ISO base media file format
E.0 Introduction This annex defines the necessary structures for the integration of AC-4 coded bitstreams in a file format that is compliant with the ISO based media file format [2]. Examples of file formats that are derived from the ISO base media file format include the MP4 file format and the 3GPP file format.
This annex additionally covers:
• the steps required to properly packetize an AC-4 bitstream for multiplexing and storage in an MPEG-DASH-compliant ISO base media file format file; and
• the steps required to demultiplex an AC-4 bitstream from an MPEG DASH-compliant ISO base media file format file.
E.1 AC-4 track definition In the terminology of the ISO base media file format specification (ISO/IEC 14496-12 [2]), AC-4 tracks are audio tracks. It therefore follows that these rules apply to the media box in the AC-4 tracks:
• In the Handler Reference Box, the handler_type field shall be set to soun.
• The Media Information Header Box shall contain a Sound Media Header Box.
• The Sample Description Box shall contain a box derived from AudioSampleEntry. This box is called AC4SampleEntry and is defined in clause E.3.
The value of the timescale parameter in the Media Header Box depends on frame_rate and base_samp_freq. The time scale shall be set according to table E.1.
NOTE: For the definition of samples, see clause E.2.
The Sample Table Box (stbl) of an AC-4 audio track shall contain a Sync Sample Box (stss), unless all samples are sync samples. The Sync Sample Box shall reference all sync samples part of that track. For Movie Fragments this corresponds to sample_is_non_sync_sample = false. The sequence_counter of the first sample should be set to 0.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)278
Table E.1: Timescale for Media Header Box
base_samp_freq [kHz]
frame_rate_index frame_rate [fps]
Media Time Scale[1/sec]
'sample_delta'[units of media time scale]
48 0 23,976 48 000 2 002 - 1 24 48 000 2 000 - 2 25 48 000 1 920 - 3 29,97 240 000 8 008 - 4 30 48 000 1 600 - 5 47,95 48 000 1 001 - 6 48 48 000 1 000 - 7 50 48 000 960 - 8 59,94 240 000 4 004 - 9 60 48 000 800 - 10 100 48 000 480 - 11 119,88 240 000 2 002 - 12 120 48 000 400 - 13 (23,44) 48 000 2 048 - 14 reserved - - - 15 reserved - - 44,1 0...12 reserved - - - 13 (21,53) 44 100 2 048 - 14, 15 reserved - -
E.2 AC-4 sample definition For the purpose of carrying AC-4 in ISO base media file format, an AC-4 sample corresponds to one raw_ac4_frame, as defined in clause 4.2.1.
NOTE: All samples in one substream share the same duration and timestamps.
Sync samples are defined as samples that have the b_iframe_global flag set to true in ac4_toc.
E.3 AC4SampleEntry Box The box type of the AC4SampleEntry Box shall be ac-4.
The AC4SampleEntry Box is defined by table E.2. Box entry values shall be set according to the values given in the table, except where left empty.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)279
Table E.2: AC4SampleEntry Box definition
Syntax No. of bits Value AC4SampleEntry() - - { - - BoxHeader.Size; 32 see note 1 BoxHeader.Type; 32 'ac-4' Reserved[6]; 8 0 DataReferenceIndex; 16 see note 1 Reserved[2]; 32 0 ChannelCount; 16 see note 2; see note 3 SampleSize; 16 16 Reserved; 32 0 SamplingFrequency; 16 see note 3 Reserved; 16 0 Ac4SpecificBox(); - - } - - NOTE 1: The values shall be set according to the sampleEntry definition in ISO/IEC 13818-1 [i.2]. NOTE 2: The ChannelCount field should be set to the total number of audio output channels of the default
presentation of that track, if not defined differently by an application standard. NOTE 3: The values of the ChannelCount and SamplingFrequency fields within the AC4SampleEntry Box shall be
ignored on decoding.
The layout of the AC4SampleEntry box is identical to that of AudioSampleEntry defined in ISO/IEC 14496-12 [2] (including the reserved fields and their values), except that AC4SampleEntry ends with a box containing AC-4 bitstream information called AC4SpecificBox. The AC4SpecificBox field structure for AC-4 is defined in clause E.4.
E.4 AC4SpecificBox The AC4SpecificBox is defined table E.3. Box entry values shall be set according to the values given in the table, except where left empty.
Table E.3: AC4SpecificBox definition
Syntax No. of bits Value AC4SpecificBox() - - { - - BoxHeader.Size; 32 - BoxHeader.Type; 32 'dac4' ac4_dsi(); - - } - -
The AC4SpecificBox() shall contain ac4_dsi() as specified in table E.4.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)280
Table E.4: AC-4 decoder specific information
Syntax No. of bits ac4_dsi() - { - ac4_dsi_version; 3 bitstream_version; 7 fs_index; 1 frame_rate_index; 4 n_presentations; 9 for (i = 0; i < n_presentations; i++) { - b_single_substream; 1 presentation_config; 5 presentation_version; 5 if (b_single_substream !=1 && presentation_config == 6) { - b_add_emdf_substreams = 1; - } else { - mdcompat; 3 if (b_belongs_to_presentation_id) { 1 presentation_id; 5 } - dsi_frame_rate_multiply_info; 2 emdf_version; 5 key_id; 10 if (b_single_substream == 1) { - ac4_substream_dsi(); - } else { - b_hsf_ext; 1 switch (presentation_config) { - case 0: - case 1: - case 2: - ac4_substream_dsi(); - ac4_substream_dsi(); - break; - case 3: - case 4: - ac4_substream_dsi(); - ac4_substream_dsi(); - ac4_substream_dsi(); - break; - case 5: - ac4_substream_dsi(); - break; - default: - n_skip_bytes; 7 n_skip_bits = n_skip_bytes * 8; - skip_bits; n_skip_bits break; - } - b_pre_virtualized; 1 b_add_emdf_substreams; 1 } - } - if (b_add_emdf_substreams) { - n_add_emdf_substreams; 7 for (j = 0; j < n_add_emdf_substreams; j++) { - emdf_version; 5 key_id; 10 } - } - byte_align; 0...7 } - } - NOTE: The number of bits in byte_align shall pad the number of bits, counted from the start of ac4_dsi to an
integer number of bytes.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)281
Table E.5: AC-4 substream decoder specific information
Syntax No. of bits ac4_substream_dsi() - { - channel_mode; 5 dsi_sf_multiplier; 2 if (b_bitrate_indicator) { 1 bitrate_indicator; 5 } - if (ch_mode >= 7 && ch_mode <= 10) { - add_ch_base; 1 } - if (b_content_type) { 1 content_classifier; 3 if (b_language_indicator) { 1 n_language_tag_bytes; 6 for (i = 0; i < n_language_tag_bytes; i++) { - language_tag_bytes; 8 } - } - } - } -
Semantics:
ac4_dsi_version - 3 bits
This field indicates the version of the decoder-specific information. For a decoder-specific information (DSI) that conforms to the present document, the ac4_dsi_version field shall be set to 000.
bitstream_version - 7 bits
This field shall contain the bitstream version as described in clause 4.3.3.2.1. Its value shall be the same as read from ac4_toc.
fs_index - 1 bit
This field shall contain the sampling frequency index as described in clause 4.3.3.2.5. Its value shall be the same as read from ac4_toc.
frame_rate_index - 4 bits
This field shall contain the frame rate index as described in clause 4.3.3.2.6. Its value shall be the same as read from ac4_toc.
n_presentations - 9 bits
This field shall contain the number of presentations contained in the corresponding AC-4 frame. Its value shall be the same as read from ac4_toc.
b_single_substream - 1 bit
This Boolean indicates that the presentation contains a single substream. Its value shall be the same as the respective value from the respective ac4_presentation_info element.
presentation_config - 5 bits
If b_single_substream is false, this field shall contain the presentation configuration as described in clause 4.3.3.3.4 and its value shall be the same as the respective value read from the respective ac4_presentation_info element. If the b_single_substream element is true the presentation_config element value should have a default value of 0.
presentation_version - 5 bits
This field shall contain the presentation version as described in clause 4.3.3.4. Its value shall be the same as the respective value read from the respective ac4_presentation_info element.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)282
mdcompat - 3 bits
This field contains the decoder compatibility indication as described in clause 4.3.3.3.8. Its value shall be the same as the respective value read from the respective ac4_presentation_info element.
b_belongs_to_presentation_id - 1 bit
This Boolean indicates that the containing presentation belongs to a presentation identifier. Its value shall be the same as the respective value from the respective ac4_presentation_info element.
presentation_id - 5 bits
This field shall contain a presentation identifier. Its value shall be the same as the respective value read from the respective ac4_presentation_info element.
dsi_frame_rate_multiply_info - 2 bits
This field shall signal the frame_rate_multiply_info element as described in clause 4.3.3.5. Its value shall correspond to the respective value read from the respective ac4_presentation_inf element as follows:
frame_rate_index b_multiplier multiplier_bit dsi_frame_rate_multiply_info 2, 3, 4 0 X 00
1 0 01 1 1 10
0, 1, 7, 8, 9 0 X 00 1 X 01
5, 6, 10, 11, 12, 13 X X 00
emdf_version - 5 bits
This field shall contain the EMDF syntax version as described in clause 4.3.3.6.1. Its value shall be the same as the respective value read from the emdf_info field in the respective ac4_presentation_info element.
key_id - 10 bits
This field shall contain the authentication ID as described in clause 4.3.3.6.2. Its value shall be the same as the respective value read from the emdf_info field in the respective ac4_presentation_info element.
b_hsf_ext - 1 bit
This Boolean shall indicate the availability of spectral data for high sampling frequencies as described in clause 4.3.3.3.3. Its value shall be the same as the respective value read from the respective ac4_presentation_info element.
n_skip_bytes - 7 bits
This field indicates a number of subsequent bytes to skip.
skip_bits
This field indicates n_skip_bytes × 8 bits to skip.
b_pre_virtualized - 1 bit
This Boolean indicates pre-rendering as described in clause 4.3.3.3.5. Its value shall be the same as the respective value read from the respective ac4_presentation_info element.
b_add_emdf_substreams - 1 bit
This Boolean indicates the presence of additional EMDF containers as described in clause 4.3.3.3.6. Its value shall be the same as the respective value read from the respective ac4_presentation_info element.
n_add_emdf_substreams - 7 bits
This field indicates the number of additional EMDF containers as described in clause 4.3.3.3.7. Its value shall be the same as the respective value read from the respective ac4_presentation_info element.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)283
emdf_version - 5 bits
This field shall contain the EMDF syntax version as described in clause 4.3.3.6.1. Its value shall be the same as the respective value read from the emdf_info field in the respective n_add_emdf_substreams() loop in the respective ac4_presentation_info element.
key_id - 10 bits
This field shall contain the authentication ID as described in clause 4.3.3.6.2. Its value shall be the same as the respective value read from the emdf_info field in the respective n_add_emdf_substreams() loop in the respective ac4_presentation_info element.
byte_align - 0 to 7 bits
This bit is used for the byte alignment of each presentation within the ac4_dsi element. Byte alignment is defined relative to the start of the enclosing syntactic element.
channel_mode - 5 bits
This field shall contain the channel mode as described in clause 4.3.3.7.1. Its value shall correspond to the respective value read from the respective ac4_substream_info element in the respective ac4_presentation_info element and is expressed either through the ch_mode parameter from table 88 (if the channel_mode bitfield is not 1111111) or through the value 12 + variable_bits(2) (if the channel_mode bitfield is 1111111).
dsi_sf_multiplier - 2 bits
This field shall signal the sf_multiplier as described in clause 4.3.3.7.3. Its value shall correspond to the respective value read from the respective ac4_substream_info element in the respective ac4_presentation_info element as follows:
Sampling frequency b_sf_multiplier sf_multiplier dsi_sf_multiplier 48 kHz 0 - 00 96 kHz 1 0 01 192 kHz - 1 10
b_bitrate_indicator - 1 bit
This Boolean indicates presence of the bitrate_indicator as described in clause 4.3.3.7.5.
bitrate_indicator - 5 bits
This field shall contain a bit-rate indication as described in clause 4.3.3.7.5. The value shall correspond to the respective value read from the respective ac4_substream_info element in the respective ac4_presentation_info element and is expressed through the brate_ind parameter from table 90.
add_ch_base - 1 bit
This bit shall contain the additional channels coupling base as described in clause 4.3.3.7.6. This field is present only if the ch_mode value according to table 88 is in the range [7...10].
b_content_type - 1 bit
This Boolean indicates the presence of content_type information as described in clause 4.3.3.7.7.
content_classifier - 3 bits
This field shall contain the content classifier as described in clause 4.3.3.8.1. The value shall correspond to the respective value read from the respective content_type field in the respective ac4_substream_info element in the respective ac4_presentation_info element.
b_language_indicator - 1 bit
This Boolean indicates presence of programme language indication as described in clause 4.3.3.8.2.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)284
n_language_tag_bytes - 6 bits
This field shall contain the number of subsequent language tags bytes as described in clause 4.3.3.8.6.
language_tag_bytes - 8 bits
The sequence of langage_tag_bytes shall contain a language tag as described in clause 4.3.3.8.7. For the respective ac4_substream_info element in the respective ac4_presentation_info element, these values shall correspond to:
• the values of the respective language_tag_bytes values in the content_type field of the respective ac4_substream_info element in the respective ac4_presentation_info element if b_serialized_language_tag is false;
• the concatenation of language_tag_chunk fields in the content_type field of the respective ac4_substream_info element in the respective ac4_presentation_info from consecutive frames if b_serialized_language_tag is true.
The ac4_dsi shall not be used to configure the AC-4 decoder. The AC-4 decoder shall obtain its configuration only from ac4_toc, which is part of every sample.
E.5 AC-4 audio tracks in fragmented isomedia files The first AC-4 sample in a movie fragment shall be a sync sample.
The Track Fragment Header Box (tfhd) should set the default-sample-duration-present flag and provide the (constant) sample duration, as given in the sample_duration field of table E.1.
The Track Fragment Run Box (trun) shall set either the first-sample-flags-present or sample-flags-present flag unless all samples in the Track Fragment Run Box are sync samples.
The sample_flag of the Track Fragment Run Box (trun) shall set the sample_is_non_sync_sample bit to 0 for all sync samples, and 1 otherwise.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)285
Annex F (normative): AC-4 Transport in MPEG-DASH
F.0 Introduction This annex describes the requirements and recommendations for delivering AC-4 streams using the MPEG Dynamic Adaptive Streaming over HTTP (DASH) standard in conjunction with the ISO base media file format, specifically referencing AC-4 audio streams within the MPEG-DASH media presentation description (MPD) file.
F.1 Media Presentation Description (MPD)
F.1.1 Overview The MPD is an XML document. The DASH client uses the information in the MPD for constructing the HTTP URLs that then allow it to access segments containing the actual audio and video content.
F.1.2 General MPD requirements relating to AC-4
F.1.2.0 Introduction
Although the syntax of the DASH media presentation description is capable of using common XML elements to describe almost any media format, the encoding type and the configuration of an AC-4 elementary stream that is part of a content presentation constrains the parameter values of some of these elements. This clause defines the values that enable an MPD to properly describe an AC-4 elementary stream.
The MPD formats described here support the following scenarios:
• Media presentations that consist of a single AC-4 elementary stream.
• Media presentations that consist of multiple AC-4 elementary streams, with each elementary stream stored in a separate MP4 file or segment file.
It is possible for the MPD to describe multiple audio services delivered using one or more multiple AC-4 elementary streams (for example, a main audio service and an associated audio service that are intended to be decoded and then mixed together).
F.1.2.1 Adaptation sets
An adaptation set describes the overall media presentation. The adaptation set typically consists of multiple instances (representations) of the same audio, video, or audio/video content, with each instance encoded at a different data rate. A DASH representation describes the parameters of each individual encoding of an adaptation set as follows:
• The codecs attribute is required. It specifies the codecs used to encode all representations within the adaptation set. For AC-4 elementary streams, the value element of the codecs attribute shall be created according to the syntax described in IETF RFC 6381 [6]. The value consists of the dot-separated list of the four following parts of which the latter three are represented by two-digit hexadecimal numbers:
- the fourCC "ac-4"
- the bitstream_version as indicated in the ac4_dsi()
- the presentation_version as indicated for the presentation in the ac4_dsi()
- the mdcompat parameter as indicated for the presentation in the ac4_dsi()
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)286
NOTE: The values in the @codecs parameter are zero-based.
• The mimeType attribute describes the encapsulation format used to store the AC-4 elementary streams present in the adaptation set. For adaptation sets that conform to ISO/IEC 14496-12 [2], the type element of the mimeType attribute shall be set to one of the following values:
- audio/mp4 (for ISO base media files that contain an AC-4 audio track but no accompanying video track);
- video/mp4 (for ISO base media files that contain AC-4 audio tracks and one or more video tracks).
Players implementing AC-4 playback compliant with the present specification shall support a codec indication of codecs="ac-4.00.00.0x, 0≤x≤n where n is the compatibility level of the decoder as defined in clause 4.3.3.3.8.
In some applications, multiple AC-4 elementary streams may be used to simultaneously deliver different audio elements of the overall media presentation. For example, one elementary stream carries a main audio service (the main audio), and a second elementary stream carries an associated audio service (such as Commentary) intended to be mixed with the main audio service before presentation to the listener.
If the content provider wishes to enable user-defined selection of specific combinations of elementary streams in the playback device (allowing different renditions of the overall media presentation to be selected and delivered), separate adaptation sets may be defined for each elementary stream. For example, one adaptation set is used to describe the main audio service on its own, and a second adaptation set describes the associated audio service that is simultaneously delivered with the main audio service to the playback device, where both adaptation sets are decoded and mixed together. Refer to clause F.1.3.0 for more details.
F.1.2.2 Representations
Each adaptation set carries one or more representations. All representations in an adaptation set shall be perceptually identical, meaning that the bit rate is the only major parameter that may differ across the AC-4 elementary streams in one adaptation set.
F.1.2.3 AudioChannelConfiguration descriptor
The representation DASH element shall include an AudioChannelConfiguration DASH descriptor, which unambiguously describes the channel configuration of the referenced AC-4 elementary stream. Refer to clause F.1.4.1 for details.
F.1.2.4 Accessibility descriptor
If the adaptation set provides for enhanced accessibility, the AdaptationSet may include an accessibility descriptor that describes the type of accessible audio service being provided. The required attribute schemeIdUri should be set to urn:tva:metadata:cs:AudioPurposeCS:2007, as defined in clause B.1 of ETSI TS 102 822-3-1 [3], signalling the namespace for the accessibility descriptor.
The audio purpose classification scheme (AudioPurposeCS), which is used to describe the type of accessible audio service that is being delivered, is defined in clause A.15 of ETSI TS 102 822-3-1 [3]. The value of the termID attribute should be set to match the type of accessible audio service carried in the AC-4 elementary stream, which is indicated by the value of the content_classifier parameter in the default presentation of the AC-4 stream, or in the AC4SpecificBox of the AC4 audio track. The corresponding values of the termID attribute and content_classifier parameter are listed in table F.1.
Table F.1: Corresponding termID attribute and content_classifier parameter values
termID attribute value AudioPurposeCS name content_classifier parameter value 1 Audio description for the visually impaired 010 2 Audio description for the hearing impaired 011 3 Supplemental commentary 101 4 Director's commentary 101 5 Educational notes 101 6 Main programme audio 000 7 Clean feed (no effects mix) 100
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)287
F.1.3 MPD with associated audio services using AC-4
F.1.3.0 Introduction
It is useful in some scenarios to simultaneously deliver two audio services (one main and one associated) to a decoder. This can be achieved as follows:
• The main audio service is a self-contained presentation that can be decoded on its own.
• The associated audio service contains supplementary audio programme elements intended to be decoded and mixed with the main audio service (for example, a director's commentary or a description of the programme for a visually impaired listener).
The main and associated audio streams are stored in separate ISO base media files that are described in separate adaptation sets within the MPD. The accessibility and role descriptors describe the purpose of the audio streams.
F.1.3.1 Role descriptor
As defined in the MPEG-DASH role scheme (urn:mpeg:dash:role:2011), the value attribute of the role descriptor shall be set to describe the purpose of each adaptation set in the overall presentation, as follows:
• If the adaptation set is delivering a full audio service intended for direct presentation to the listener, the value attribute shall be main or alternate.
• If the adaptation set is delivering an audio service intended to be mixed with a full audio service delivered in a different adaptation set before presentation to the listener (sometimes referred to as a receiver-mix service), the value attribute shall be set to commentary.
• If the adaptation set is delivering a full audio service intended for direct presentation to the listener, but this audio service is intended as an alternative presentation to the main audio service, the value attribute shall be set to alternate. This is applicable, for example, when delivering a service that contains premixed main audio and audio elements for visually impaired listeners, sometimes referred to as a broadcast-mix service.
F.1.3.2 dependencyID
An adaptation set that is delivering an associated audio service shall not be decoded and presented to the listener on its own, but shall always be mixed with the decoded audio from the adaptation set that is delivering the corresponding main audio service. Therefore, the adaptation set that is delivering the associated audio service should include a dependencyID descriptor. This descriptor indicates the relationship of the associated audio service with the main audio service that it will be mixed with after decoding.
F.1.4 Descriptors specific to AC-4 elementary streams
F.1.4.0 Introduction
The following descriptors are specific to AC-4 elementary streams.
F.1.4.1 AudioChannelConfiguration descriptor
For AC-4 elementary streams, the AudioChannelConfiguration descriptor shall use the AudioChannelConfiguration scheme described in the schemeIdUri tag:dolby.com,2014:dash:audio_channel_configuration:2011.
The value element, if used, shall contain a four-digit hexadecimal representation of the 16-bit bit field, which describes the channel assignment of the referenced AC-4 elementary stream according to table F.2.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)288
Table F.2: AudioChannelConfiguration descriptor
Bit Speaker location 0 (MSB, see note 2) L 1 C 2 R 3 Ls 4 Rs 5 Lc/Rc pair 6 Lrs/Rrs pair 7 Cs 8 Ts 9 Lsd/Rsd pair 10 Lw/Rw pair 11 Vhl/Vhr pair 12 Vhc 13 LFE2 14 LFE 15 Reserved NOTE 1: See table D.1 for an explanation of speaker location acronyms. NOTE 2: Bit 0, which indicates the presence of the L channel, is the MSB of the AudioChannelConfiguration
descriptor. For example, to indicate that the channel configuration of the AC-4 elementary stream is L, C, R, Ls, Rs, LFE, the value element would contain the value F802 (the hexadecimal equivalent of the binary value 0b1111100000000010).
.
F.1.5 MPD manifest file examples
F.1.5.0 Introduction
Clause F.1.5 contains example MPD files for different media presentations.
F.1.5.1 MPD for a single video component and single audio component
The following MPD example describes a simple media presentation that consists of a single video component with a single 5.1-channel (L, C, R, Ls, Rs, LFE) AC-4 audio component. Three representations of the video content and three representations of the audio content are provided, each at a different data rate.
The media presentation complies with the ISO base media file format live profile, as defined in ISO/IEC 23009-1 [1].
Pseudocode 124
<?xml version="1.0" encoding="UTF-8"?> <MPD xmlns="urn:mpeg:dash:schema:mpd:2011" xmlns:dolby="http://www.dolby.com/ns/online/DASH" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="urn:mpeg:dash:schema:mpd:2011 http://standards.iso.org/ittf/PubliclyAvailableStandards/MPEG-DASH_schema_files/DASH-MPD.xsd" type="static" minimumUpdatePeriod="PT2S" timeShiftBufferDepth="PT30M" availabilityStartTime="2011-12-25T12:30:01" minBufferTime="PT4S" profiles="urn:mpeg:dash:profile:isoff-live:2011"> <BaseURL>http://cdn1.example.com/</BaseURL> <BaseURL>http://cdn2.example.com/</BaseURL> <Period> <!-- Video --> <AdaptationSet mimeType="video/mp4" frameRate="30000/1001" segmentAlignment="true" startWithSAP="1"> <BaseURL>video/</BaseURL> <SegmentTemplate timescale="90000" media="$Bandwidth$/$Number$.m4s" initialization="$Bandwidth$/0.mp4"> <SegmentTimeline>
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)289
<S t="0" d="180180" r="10"/> </SegmentTimeline> </SegmentTemplate> <Representation id="v0" width="320" height="240" bandwidth="250000" codecs="avc1.4D401F" /> <Representation id="v1" width="640" height="480" bandwidth="500000" codecs="avc1.4D401F" /> <Representation id="v2" width="960" height="720" bandwidth="1000000" codecs="avc1.4D401F" /> </AdaptationSet> <!-- 5.1 channel English Audio --> <AdaptationSet mimeType="audio/mp4" codecs="ac-4.00.00.03" lang="en" segmentAlignment="true" startWithSAP="1"> <Role schemeIdUri="urn:mpeg:dash:role:2011" value="main"/> <SegmentTemplate timescale="48000" media="audio/en/$Bandwidth$/$Number$.m4s" initialization="audio/en/$Bandwidth$/0.mp4"> <SegmentTimeline> <S t="0" d="96768" r="10"/> </SegmentTimeline> </SegmentTemplate> <Representation id="a0" bandwidth="192000" > <AudioChannelConfiguration schemeIdUri="tag:dolby.com,2014:dash:audio_channel_configuration:2011" value="F802"/> </Representation> <Representation id="a1" bandwidth="256000" > <AudioChannelConfiguration schemeIdUri="tag:dolby.com,2014:dash:audio_channel_configuration:2011" value="F802"/> </Representation> <Representation id="a2" bandwidth="384000" > <AudioChannelConfiguration schemeIdUri="tag:dolby.com,2014:dash:audio_channel_configuration:2011" value="F802"/> </Representation> </AdaptationSet> </Period> </MPD>
F.1.5.2 MPD for main and associated audio services delivered in separate files
This is a simple example of a static presentation, with multiple languages and multiple base URLs. The following MPD document describes content available from two sources (cdn1 and cdn2) with audio available in two different English language presentations: main audio service only or a visually impaired receiver-mix service. The visually impaired service is enabled by simultaneously delivering the AC-4 bitstream containing the main audio service and an additional AC-4 bitstream containing the associated audio service for visually impaired listeners. The AC-4 streams are coded at the same frame rate as video, enabling seamless and easy switching and trick play.
Three versions of the video are provided at bit rates between 250 kbps and 1 Mbps in different spatial resolutions.
The media presentation complies with the ISO base media file format live profile, as defined in ISO/IEC 23009-1 [1].
Pseudocode 125
<?xml version="1.0" encoding="utf-8"?> <MPD xmlns="urn:mpeg:dash:schema:mpd:2011" xmlns:dolby="http://www.dolby.com/ns/online/DASH" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="urn:mpeg:dash:schema:mpd:2011 http://standards.iso.org/ittf/PubliclyAvailableStandards/MPEG-DASH_schema_files/DASH-MPD.xsd" type="static" minimumUpdatePeriod="PT2S" timeShiftBufferDepth="PT30M" availabilityStartTime="2011-12-25T12:30:00" minBufferTime="PT4S" profiles="urn:mpeg:dash:profile:isoff-live:2011"> <BaseURL>http://cdn1.example.com/</BaseURL> <BaseURL>http://cdn2.example.com/</BaseURL> <Period> <!-- Video --> <AdaptationSet mimeType="video/mp4" frameRate="30000/1001" segmentAlignment="true" startWithSAP="1"> <BaseURL>video/</BaseURL> <SegmentTemplate timescale="90000" media="$Bandwidth$/$Number$.m4s" initialization="$Bandwidth$/0.mp4">
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)290
<SegmentTimeline > <S t="0" d="180180" r="12"/> </SegmentTimeline> </SegmentTemplate> <Representation id="v0" width="320" height="240" bandwidth="250000" codecs="avc1.4D401F" /> <Representation id="v1" width="640" height="480" bandwidth="500000" codecs="avc1.4D401F" /> <Representation id="v2" width="960" height="720" bandwidth="1000000" codecs="avc1.4D401F" /> </AdaptationSet> <!-- English Audio --> <AdaptationSet mimeType="audio/mp4" codecs="ac-4.00.00.03" lang="en" segmentAlignment="0" startWithSAP="1"> <Role schemeIdUri="urn:mpeg:dash:role:2011" value="main" /> <SegmentTemplate timescale="48000" media="audio/en_main/$Bandwidth$/$Number$.m4s" initialization="audio/en_main/$Bandwidth$/0.mp4"> <SegmentTimeline> <S t="0" d="96000" r="11"/> </SegmentTimeline> </SegmentTemplate> <Representation id="a0" bandwidth="256000"> <AudioChannelConfiguration schemeIdUri="tag:dolby.com,2014:dash:audio_channel_configuration:2011" value="F802"/> </Representation> </AdaptationSet> <!-- English Audio for visually impaired listeners --> <AdaptationSet mimeType="audio/mp4" codecs="ac-4.00.00.03" lang="en" segmentAlignment="true" startWithSAP="1"> <!-- Audio description for the visually impaired --> <Accessibility schemeIdUri="urn:tva:metadata:cs:AudioPurposeCS:2007" value="1"/> <Role schemeIdUri="urn:mpeg:dash:role:2011" value="commentary" /> <SegmentTemplate timescale="48000" media="audio/en_vi/$Bandwidth$/$Number$.m4s" initialization="audio/en_vi/$Bandwidth$/0.mp4"> <SegmentTimeline> <S t="0" d="96000" r="11"/> </SegmentTimeline> </SegmentTemplate> <Representation id="a1" dependencyId="a0" bandwidth="64000"> <AudioChannelConfiguration schemeIdUri="tag:dolby.com,2014:dash:audio_channel_configuration:2011" value="4000"/> </Representation> </AdaptationSet> <!-- French Audio dub --> <AdaptationSet mimeType="audio/mp4" codecs="ac-4.00.00.03" lang="fr" segmentAlignment="0" startWithSAP="1"> <!-- If french was original content (as in a live french commentary), then the role @value could be main as well --> <Role schemeIdUri="urn:mpeg:dash:role:2011" value="dub"/> <SegmentTemplate timescale="48000" media="audio/fr/$Bandwidth$/$Number$.m4s" initialization="audio/fr/$Bandwidth$/0.mp4"> <SegmentTimeline> <S t="0" d="96000" r="11"/> </SegmentTimeline> </SegmentTemplate> <Representation id="a2" bandwidth="192000"> <AudioChannelConfiguration schemeIdUri="tag:dolby.com,2014:dash:audio_channel_configuration:2011" value="F802"/> </Representation> </AdaptationSet> </Period> </MPD>
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)291
Annex G (normative): AC-4 Sync Frame
G.1 Introduction The AC-4 sync frame is an optional bitstream layer for encapsulating AC-4 raw frames. It can be used when the transmission system is not frame based, i.e. when it does not include information about the framing. In this case the AC-4 sync frame offers a simple way to reconstruct the framing at the receiver side by a high level parsing of the bitstream.
The sync_word can be either 0xAC40 or 0xAC41. If the sync_word is 0xAC41, a crc_word is also transmitted.
The crc_word is a 16 bit CRC based on the frame_size element and the raw_ac4_frame element (the sync_word element is not part of the CRC calculation). The following generator polynomial is used to generate each of the 16-bit CRC words: x16 + x15 + x2 + 1. The polynomial is also commonly known as IBM-CRC-16.
G.2 ac4_syncframe Syntax No of bits
ac4_syncframe() { sync_word; .................................................................................. 16 frame_size(); raw_ac4_frame(); if (sync_word == 0xAC41) { crc_word; ................................................................................. 16 } }
G.3 frame_size Syntax No of bits
frame_size() { frame_size; ................................................................................. 16 if (frame_size == 0xffff) { frame_size; ............................................................................... 24 } }
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)292
Annex H (informative): Bibliography
• Charles Q. Robinson, Kenneth Gundry: "Dynamic Range Control via Metadata", Preprint 5028, 107th AES Convention New York, Sept. 1999.
• Davidson, G. A.: "The Digital Signal Processing Handbook", Madisetti, V. K. and Williams, D. B. Eds.(CRC Press LLC, 1997), pp. 41-1 - 41-21.
• Princen J., Bradley A.: "Analysis/synthesis filter bank design based on time domain aliasing cancellation", IEEE Trans. Acoust. Speech and Signal Processing, vol. ASSP-34, pp. 1153-1161, Oct. 1986.
• R. Rao, P. Yip: "Discrete Cosine Transform", Academic Press, Boston 1990, pp. 11.
• Cover, T. M., Thomas, J. A.: "Elements of Information Theory", Wiley Series in Telecommunications, New York, 1991, pp. 13.
• Gersho A., Gray R. M.: "Vector Quantization and Signal Compression", Kluwer Academic Publisher, Boston, 1992, pp. 309.
• A. Moffat, R. Neal and I. H. Witten: "Arithmetic Coding Revisited", in Proc. IEEE Data Compression Conference, Snowbird, Utah, March 1995.
ETSI
ETSI TS 103 190-1 V1.3.1 (2018-02)293
History
Document history
V1.1.1 April 2014 Publication as ETSI TS 103 190
V1.2.1 June 2015 Publication
V1.3.1 February 2018 Publication